{"id": "c-darwin-01-dessert-01", "uid": "c000", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "prompt_sha256_by_order": ["ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d"], "request_sha256": "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "compiled_prompt_sha256_by_order": ["2ca65940a5765f77af59cf411c0b326a074bbc7804d92041b9a96860a2ba21f0", "48dbee6ba046a0f5d1492a021834e80774f5e6d672ab585b30be1393344d7180", "8469ceacf69d457d0646180caed55919f309bd99e78602af7c4c31eca9d50741", "218d7301ee961a1bc56ec764ba057f602e68d84c1990b15df5371a4b82fa233e", "f13adde1c9aaa8a22e12b94a59507cb89bd9e3cf532778ae7d31880236e63212", "3c31ef585bd4ffd3fc8a990fd1ab1073f330e1883dd469671aa08c5027d3084d"], "compiled_tokens_by_order": [279, 279, 279, 279, 279, 279], "choice": "sagan", "probs": {"darwin": 0.34613754380750206, "einstein": 0.04684462253001373, "hypatia": 0.019527743958938876, "ibn_sina": 0.008140374786488052, "sagan": 0.5706843310633261, "socrates": 0.008665383853731185}, "confidence": 0.44864898093302397, "seconds": 0.17011356100010744, "seconds_all_orders": 1.0206813660006446, "seconds_by_order": [0.1945128970000951, 0.16477582500010612, 0.16467300600197632, 0.16471197600185405, 0.1645509369991487, 0.16745672499746433], "argmax_stable": false, "spread": {"darwin": 0.3924218162501799, "einstein": 0.3029923493333471, "hypatia": 0.014709520350820128, "ibn_sina": 0.01255573906651139, "sagan": 0.6118171897685811, "socrates": 0.021842020025072503}, "spread_2dp": {"darwin": 0.38999999999999996, "einstein": 0.3, "hypatia": 0.02, "ibn_sina": 0.02, "sagan": 0.6100000000000001, "socrates": 0.019999999999999997}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "einstein", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.34613754380750206, "einstein": 0.04684462253001373, "hypatia": 0.019527743958938876, "ibn_sina": 0.008140374786488052, "sagan": 0.5706843310633261, "socrates": 0.008665383853731185}, {"ibn_sina": 0.014841173539581996, "hypatia": 0.004818223608118748, "sagan": 0.7612069868412102, "socrates": 0.01309728967940157, "darwin": 0.10301816316584371, "einstein": 0.10301816316584371}, {"hypatia": 0.00856352100888435, "darwin": 0.02054281897758086, "sagan": 0.9298389865285481, "ibn_sina": 0.010995778631746955, "einstein": 0.01502944742661983, "socrates": 0.01502944742661983}, {"ibn_sina": 0.01743923801363278, "einstein": 0.039299929526212446, "darwin": 0.22615547961837876, "sagan": 0.6966079187956686, "hypatia": 0.012758819087694129, "socrates": 0.007738614958413261}, {"einstein": 0.31802179675996695, "sagan": 0.31802179675996695, "socrates": 0.029580634983485763, "ibn_sina": 0.010882107467221158, "darwin": 0.31802179675996695, "hypatia": 0.005471867269392175}, {"socrates": 0.010338367352183967, "ibn_sina": 0.004883498947121393, "hypatia": 0.009711997338805983, "einstein": 0.03184441330850115, "sagan": 0.5302570878256267, "darwin": 0.4129646352277608}], "prompt_tokens": 279, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.34613754380750206, "einstein": 0.04684462253001373, "hypatia": 0.019527743958938876, "ibn_sina": 0.008140374786488052, "sagan": 0.5706843310633261, "socrates": 0.008665383853731185}, "confidence": 0.44864898093302397, "choice": "sagan"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 194.446606998099, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "2ca65940a5765f77af59cf411c0b326a074bbc7804d92041b9a96860a2ba21f0", "compiled_ids_sha256": "39875e59687c95458e3c8bef14fa71362355147df3bbbb6895cd7ad2bb135fb8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 16.5, 15.625, 14.75, 19.0, 14.8125], "runtime_probabilities": [0.34613755345344543, 0.04684462398290634, 0.019527744501829147, 0.008140374906361103, 0.5706843733787537, 0.008665383793413639], "decide_seconds": 0.1874971600009303, "probabilities_f64_minus_runtime_max_abs": 4.2315427606354206e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.014841173539581996, "hypatia": 0.004818223608118748, "sagan": 0.7612069868412102, "socrates": 0.01309728967940157, "darwin": 0.10301816316584371, "einstein": 0.10301816316584371}, "confidence": 0.541813648094905, "choice": "sagan"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 164.75262500171084, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "48dbee6ba046a0f5d1492a021834e80774f5e6d672ab585b30be1393344d7180", "compiled_ids_sha256": "1c43b23c4ff7ee2b0043a4d253a4bbf977882d588e01e9dacd72a1c7b6474d65", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 14.1875, 19.25, 15.1875, 17.25, 17.25], "runtime_probabilities": [0.014841174706816673, 0.004818223882466555, 0.7612070441246033, 0.013097289949655533, 0.10301816463470459, 0.10301816463470459], "decide_seconds": 0.16122311200160766, "probabilities_f64_minus_runtime_max_abs": 5.7283393029017304e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.00856352100888435, "darwin": 0.02054281897758086, "sagan": 0.9298389865285481, "ibn_sina": 0.010995778631746955, "einstein": 0.01502944742661983, "socrates": 0.01502944742661983}, "confidence": 0.7968524470670013, "choice": "sagan"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 164.65013600100065, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "8469ceacf69d457d0646180caed55919f309bd99e78602af7c4c31eca9d50741", "compiled_ids_sha256": "b850e58e668e249c6ab16ac15630067755df76d9a0cbf85e7e4b392802f09665", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 15.8125, 19.625, 15.1875, 15.5, 15.5], "runtime_probabilities": [0.008563520386815071, 0.020542817190289497, 0.9298389554023743, 0.010995777323842049, 0.015029446221888065, 0.015029446221888065], "decide_seconds": 0.16125420200114604, "probabilities_f64_minus_runtime_max_abs": 3.112617386058503e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01743923801363278, "einstein": 0.039299929526212446, "darwin": 0.22615547961837876, "sagan": 0.6966079187956686, "hypatia": 0.012758819087694129, "socrates": 0.007738614958413261}, "confidence": 0.5093588094662796, "choice": "sagan"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 164.68940599952475, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "218d7301ee961a1bc56ec764ba057f602e68d84c1990b15df5371a4b82fa233e", "compiled_ids_sha256": "03f544c4c81ba324ffa4af0f748ddef4ec3b4480551eb9123c4a9b91dd0308a9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 16.25, 18.0, 19.125, 15.125, 14.625], "runtime_probabilities": [0.017439238727092743, 0.03929993137717247, 0.22615547478199005, 0.6966078877449036, 0.01275881938636303, 0.00773861538618803], "decide_seconds": 0.16128393200051505, "probabilities_f64_minus_runtime_max_abs": 3.105076507026183e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.31802179675996695, "sagan": 0.31802179675996695, "socrates": 0.029580634983485763, "ibn_sina": 0.010882107467221158, "darwin": 0.31802179675996695, "hypatia": 0.005471867269392175}, "confidence": 0.2884949354282442, "choice": "einstein"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 164.5282169993152, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "f13adde1c9aaa8a22e12b94a59507cb89bd9e3cf532778ae7d31880236e63212", "compiled_ids_sha256": "26ae9000f113b6924b052671c5bf5b7966533710e6a39fc0ca6ea6923d555ae4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.625, 18.625, 16.25, 15.25, 18.625, 14.5625], "runtime_probabilities": [0.3180218040943146, 0.3180218040943146, 0.029580634087324142, 0.01088210754096508, 0.3180218040943146, 0.0054718670435249805], "decide_seconds": 0.16113212299751467, "probabilities_f64_minus_runtime_max_abs": 7.334347629228688e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.010338367352183967, "ibn_sina": 0.004883498947121393, "hypatia": 0.009711997338805983, "einstein": 0.03184441330850115, "sagan": 0.5302570878256267, "darwin": 0.4129646352277608}, "confidence": 0.48115602680049085, "choice": "sagan"}}, "usage": {"input_tokens": 279, "output_tokens": 1}, "latency_ms": 167.43456499898457, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 279, "compiled_prompt_sha256": "3c31ef585bd4ffd3fc8a990fd1ab1073f330e1883dd469671aa08c5027d3084d", "compiled_ids_sha256": "ef6a744ad93767063b46b3134aab606d4679224932cb3b5550b206fd09ee53ce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 13.9375, 14.625, 15.8125, 18.625, 18.375], "runtime_probabilities": [0.010338367894291878, 0.004883498419076204, 0.009711997583508492, 0.03184441104531288, 0.5302570462226868, 0.4129646122455597], "decide_seconds": 0.16407822200199007, "probabilities_f64_minus_runtime_max_abs": 4.160293998189957e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:47:55Z"} {"id": "c-darwin-01-dessert-01", "uid": "c001", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "prompt_sha256_by_order": ["ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66"], "request_sha256": "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "compiled_prompt_sha256_by_order": ["dfdf84665a6deae6826cfe5cb74cbdf7164c3c10210051c62958077957da1655", "1b455f62cd412deb3a24d61fab9d3554f21bb2aed9d2368dd8596679da0f9778", "5868f7f26b62319d9696c2266c1d8d6f001d23651b72baa28b1386b803b7ce58", "fd499f127c17859447da2316773685477d9baac7cb8267200f21b6bfeca90015", "c104604437579117b904b7c741fcf93ef03da60367be03d47a5c692e96a90bde", "7cb668279fdd9d43553280b896c143d250d5ea31cae48d76db701d614598e31a"], "compiled_tokens_by_order": [379, 379, 379, 379, 379, 379], "choice": "darwin", "probs": {"darwin": 0.9955267194409912, "einstein": 0.0011656419477465497, "hypatia": 0.001028675408409005, "ibn_sina": 0.0007069975795554337, "sagan": 0.0009078028616858906, "socrates": 0.0006641627616118841}, "confidence": 0.9800416375701024, "seconds": 0.1715377475005274, "seconds_all_orders": 1.0292264850031643, "seconds_by_order": [0.17220433900001808, 0.17126041700248607, 0.17136702600328135, 0.1720206899990444, 0.17127858599997126, 0.1710954269983631], "argmax_stable": true, "spread": {"darwin": 0.03999138138485092, "einstein": 0.03588448384059698, "hypatia": 0.001219871476787602, "ibn_sina": 0.001909952596996624, "sagan": 0.0011858312841366153, "socrates": 0.0013077119121826387}, "spread_2dp": {"darwin": 0.040000000000000036, "einstein": 0.04, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9955267194409912, "einstein": 0.0011656419477465497, "hypatia": 0.001028675408409005, "ibn_sina": 0.0007069975795554337, "sagan": 0.0009078028616858906, "socrates": 0.0006641627616118841}, {"ibn_sina": 0.0026169501765520575, "hypatia": 0.0014007524898642782, "sagan": 0.000904393780971072, "socrates": 0.0017986017994743712, "darwin": 0.9917882084893622, "einstein": 0.001491093263775919}, {"hypatia": 0.002174160796349567, "darwin": 0.9939061138477256, "sagan": 0.000964778005063859, "ibn_sina": 0.000964778005063859, "einstein": 0.0012387994799632892, "socrates": 0.0007513698658338011}, {"ibn_sina": 0.0018978309266350695, "einstein": 0.012375412193555776, "darwin": 0.9831007590491694, "sagan": 0.0010158356928782063, "hypatia": 0.0009542893195619651, "socrates": 0.0006558728181996706}, {"einstein": 0.03705012578834353, "sagan": 0.0020902250651076873, "socrates": 0.0019635847303823093, "ibn_sina": 0.001732857442524796, "darwin": 0.9555353380561403, "hypatia": 0.0016278689175013453}, {"socrates": 0.0018001661290139085, "ibn_sina": 0.0014923901385146252, "hypatia": 0.0009635594937874855, "einstein": 0.0016910995768300342, "sagan": 0.0014019707909346515, "darwin": 0.9926508138709191}], "prompt_tokens": 379, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9955267194409912, "einstein": 0.0011656419477465497, "hypatia": 0.001028675408409005, "ibn_sina": 0.0007069975795554337, "sagan": 0.0009078028616858906, "socrates": 0.0006641627616118841}, "confidence": 0.9800416375701024, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 172.16828999880818, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "dfdf84665a6deae6826cfe5cb74cbdf7164c3c10210051c62958077957da1655", "compiled_ids_sha256": "935f8d246f67d04de94e49c25e2c59f166760a7b2d6dddf4f9fb2ac6ee141755", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 12.875, 12.75, 12.375, 12.625, 12.3125], "runtime_probabilities": [0.9955267310142517, 0.001165641937404871, 0.0010286754695698619, 0.0007069975836202502, 0.0009078028961084783, 0.0006641627987846732], "decide_seconds": 0.16745128500042483, "probabilities_f64_minus_runtime_max_abs": 1.157326046108409e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0026169501765520575, "hypatia": 0.0014007524898642782, "sagan": 0.000904393780971072, "socrates": 0.0017986017994743712, "darwin": 0.9917882084893622, "einstein": 0.001491093263775919}, "confidence": 0.9663164551585952, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 171.23547699884512, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "1b455f62cd412deb3a24d61fab9d3554f21bb2aed9d2368dd8596679da0f9778", "compiled_ids_sha256": "abf2faa76564d81fb58d134901e31cc8f085074c776142296365eda850d32e3c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.6875, 13.0625, 12.625, 13.3125, 19.625, 13.125], "runtime_probabilities": [0.0026169505435973406, 0.001400752691552043, 0.0009043939062394202, 0.0017986019374802709, 0.9917883276939392, 0.0014910934260115027], "decide_seconds": 0.16685599999982514, "probabilities_f64_minus_runtime_max_abs": 1.1920457698710862e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.002174160796349567, "darwin": 0.9939061138477256, "sagan": 0.000964778005063859, "ibn_sina": 0.000964778005063859, "einstein": 0.0012387994799632892, "socrates": 0.0007513698658338011}, "confidence": 0.9740475758127968, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 171.3416460006556, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "5868f7f26b62319d9696c2266c1d8d6f001d23651b72baa28b1386b803b7ce58", "compiled_ids_sha256": "a514ce09513e6a1603befcb5277126f6b04f732acfadf3dd2792edbd5f03f2e9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 19.625, 12.6875, 12.6875, 12.9375, 12.4375], "runtime_probabilities": [0.0021741606760770082, 0.9939060807228088, 0.0009647778933867812, 0.0009647778933867812, 0.0012387995375320315, 0.0007513698656111956], "decide_seconds": 0.16695049900226877, "probabilities_f64_minus_runtime_max_abs": 3.3124916809512683e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0018978309266350695, "einstein": 0.012375412193555776, "darwin": 0.9831007590491694, "sagan": 0.0010158356928782063, "hypatia": 0.0009542893195619651, "socrates": 0.0006558728181996706}, "confidence": 0.9433808731481108, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 171.99540100045851, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "fd499f127c17859447da2316773685477d9baac7cb8267200f21b6bfeca90015", "compiled_ids_sha256": "7642123f71f1bc74d7c3ca6c07e85bb7c21a09942791cc6afdd5183de9fa4dc6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 15.375, 19.75, 12.875, 12.8125, 12.4375], "runtime_probabilities": [0.0018978309817612171, 0.01237541250884533, 0.9831007719039917, 0.0010158356744796038, 0.0009542892803438008, 0.0006558728055097163], "decide_seconds": 0.16763030500078457, "probabilities_f64_minus_runtime_max_abs": 1.2854822317720505e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03705012578834353, "sagan": 0.0020902250651076873, "socrates": 0.0019635847303823093, "ibn_sina": 0.001732857442524796, "darwin": 0.9555353380561403, "hypatia": 0.0016278689175013453}, "confidence": 0.8815884159661393, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 171.25170600047568, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "c104604437579117b904b7c741fcf93ef03da60367be03d47a5c692e96a90bde", "compiled_ids_sha256": "d0da3fb5c81cae0502389616395e7847afb7792ce34455d715ace63571161b0d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 13.375, 13.3125, 13.1875, 19.5, 13.125], "runtime_probabilities": [0.03705012425780296, 0.0020902249962091446, 0.0019635846838355064, 0.001732857315801084, 0.9555352926254272, 0.0016278689727187157], "decide_seconds": 0.16679795000163722, "probabilities_f64_minus_runtime_max_abs": 4.543071308038549e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0018001661290139085, "ibn_sina": 0.0014923901385146252, "hypatia": 0.0009635594937874855, "einstein": 0.0016910995768300342, "sagan": 0.0014019707909346515, "darwin": 0.9926508138709191}, "confidence": 0.9692445905355479, "choice": "darwin"}}, "usage": {"input_tokens": 379, "output_tokens": 1}, "latency_ms": 171.07113699967158, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 379, "compiled_prompt_sha256": "7cb668279fdd9d43553280b896c143d250d5ea31cae48d76db701d614598e31a", "compiled_ids_sha256": "86a7c6efd2f4acf88c8f5375d7ed27508013e0000fea6565e7ae9495cf7c34c4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 13.0, 12.5625, 13.125, 12.9375, 19.5], "runtime_probabilities": [0.0018001660937443376, 0.0014923900598660111, 0.0009635594324208796, 0.0016910997219383717, 0.00140197086147964, 0.9926508069038391], "decide_seconds": 0.16677009100021678, "probabilities_f64_minus_runtime_max_abs": 6.967080024011807e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:47:56Z"} {"id": "c-darwin-01-dessert-01", "uid": "c002", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "prompt_sha256_by_order": ["6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc"], "request_sha256": "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "compiled_prompt_sha256_by_order": ["36256d9ffe4d50c3d22a8cefe2d22c38863f669314f6c78f6e3165edd736e631", "e42bd9b0cedfb08ec693aa27f8b2df6de90ca7d55e8463294125dc41439e0a37", "0d3acb839327b5ed8b9a89af5a322a76c70bad66bcaf5f2e160d4bb6c9d57850", "9b6d91cb0b3666a5057a4df12fc8247dda9ac305984b9d9ef3535630cdbab79e", "c1bb3d6c6a862369b6084f743e6e387dfa5e1afab61d21235ee28083ce7263db", "4d89b0a4d92b78766d8c6df2302113378d8eabe3158ff0cc3f5de3002ee99f8f"], "compiled_tokens_by_order": [296, 296, 296, 296, 296, 296], "choice": "sagan", "probs": {"darwin": 0.12788655467579865, "einstein": 0.23892353878299574, "hypatia": 0.016258950860860502, "ibn_sina": 0.03037571921789179, "sagan": 0.3476316976794576, "socrates": 0.23892353878299574}, "confidence": 0.16979662361970327, "seconds": 0.16605784933320442, "seconds_all_orders": 0.9963470959992264, "seconds_by_order": [0.16913958200166235, 0.1653834709977673, 0.16503738400206203, 0.16498949399829144, 0.1650562840004568, 0.16674088099898654], "argmax_stable": false, "spread": {"darwin": 0.19947028808838468, "einstein": 0.241109217905645, "hypatia": 0.016534960080052934, "ibn_sina": 0.03801514839300215, "sagan": 0.3742654393635882, "socrates": 0.24414609827479566}, "spread_2dp": {"darwin": 0.2, "einstein": 0.25, "hypatia": 0.01, "ibn_sina": 0.04000000000000001, "sagan": 0.37, "socrates": 0.24000000000000002}, "choices_by_order": ["sagan", "einstein", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.12788655467579865, "einstein": 0.23892353878299574, "hypatia": 0.016258950860860502, "ibn_sina": 0.03037571921789179, "sagan": 0.3476316976794576, "socrates": 0.23892353878299574}, {"ibn_sina": 0.06353696362570731, "hypatia": 0.008598792968279456, "sagan": 0.22176579352949666, "socrates": 0.22176579352949666, "darwin": 0.11870267544124583, "einstein": 0.365629980905774}, {"hypatia": 0.017752011650845674, "darwin": 0.033165124002982146, "sagan": 0.16842639710718257, "ibn_sina": 0.06196060884630976, "einstein": 0.31466193552036803, "socrates": 0.40403392287231177}, {"ibn_sina": 0.043542410757400354, "einstein": 0.2211266077150967, "darwin": 0.19514354638761175, "sagan": 0.19514354638761175, "hypatia": 0.023306572983166757, "socrates": 0.32173731576911263}, {"einstein": 0.3065066038826809, "sagan": 0.06035472883168703, "socrates": 0.34731748404442636, "ibn_sina": 0.06839086761089394, "darwin": 0.21065870272719794, "hypatia": 0.006771612903113822}, {"socrates": 0.1598878245975161, "ibn_sina": 0.031483779386251236, "hypatia": 0.016852052729461577, "einstein": 0.12452076300012897, "sagan": 0.4346201681952752, "darwin": 0.23263541209136682}], "prompt_tokens": 296, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.12788655467579865, "einstein": 0.23892353878299574, "hypatia": 0.016258950860860502, "ibn_sina": 0.03037571921789179, "sagan": 0.3476316976794576, "socrates": 0.23892353878299574}, "confidence": 0.16979662361970327, "choice": "sagan"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 169.11383199840202, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "36256d9ffe4d50c3d22a8cefe2d22c38863f669314f6c78f6e3165edd736e631", "compiled_ids_sha256": "bedc1945b96a6e27ccb028c42f119ba2eb225cdab520d6ff987d956978f024cd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 17.875, 15.1875, 15.8125, 18.25, 17.875], "runtime_probabilities": [0.127886563539505, 0.2389235496520996, 0.016258951276540756, 0.03037571907043457, 0.34763169288635254, 0.2389235496520996], "decide_seconds": 0.16542847100208746, "probabilities_f64_minus_runtime_max_abs": 1.0869103866939511e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.06353696362570731, "hypatia": 0.008598792968279456, "sagan": 0.22176579352949666, "socrates": 0.22176579352949666, "darwin": 0.11870267544124583, "einstein": 0.365629980905774}, "confidence": 0.16011332585621152, "choice": "einstein"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 165.35966100127553, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "e42bd9b0cedfb08ec693aa27f8b2df6de90ca7d55e8463294125dc41439e0a37", "compiled_ids_sha256": "56ee90834bdc5b1bb8decbdf7a4904d5212754d2d6e8373e652dfb8e56f1e39e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 14.625, 17.875, 17.875, 17.25, 18.375], "runtime_probabilities": [0.06353696435689926, 0.008598793298006058, 0.22176580131053925, 0.22176580131053925, 0.11870267987251282, 0.36563000082969666], "decide_seconds": 0.16168988000208628, "probabilities_f64_minus_runtime_max_abs": 1.992392267835541e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.017752011650845674, "darwin": 0.033165124002982146, "sagan": 0.16842639710718257, "ibn_sina": 0.06196060884630976, "einstein": 0.31466193552036803, "socrates": 0.40403392287231177}, "confidence": 0.22597925100410077, "choice": "socrates"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 165.01572400011355, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "0d3acb839327b5ed8b9a89af5a322a76c70bad66bcaf5f2e160d4bb6c9d57850", "compiled_ids_sha256": "4f983f973454a9880cc52fae24d50b5d2c6b77c1652e37e2e9f1141f3d4860ea", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 16.125, 17.75, 16.75, 18.375, 18.625], "runtime_probabilities": [0.017752012237906456, 0.03316512331366539, 0.16842639446258545, 0.0619606114923954, 0.3146619200706482, 0.40403392910957336], "decide_seconds": 0.16152812099971925, "probabilities_f64_minus_runtime_max_abs": 1.544971983680199e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.043542410757400354, "einstein": 0.2211266077150967, "darwin": 0.19514354638761175, "sagan": 0.19514354638761175, "hypatia": 0.023306572983166757, "socrates": 0.32173731576911263}, "confidence": 0.1291518294804117, "choice": "socrates"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 164.96746399934636, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "9b6d91cb0b3666a5057a4df12fc8247dda9ac305984b9d9ef3535630cdbab79e", "compiled_ids_sha256": "269ac0cb3bb59fb13b4a76c3d78878c17e480f4302a5ee573041686c723ae9ed", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 17.875, 17.75, 17.75, 15.625, 18.25], "runtime_probabilities": [0.04354241117835045, 0.22112661600112915, 0.19514355063438416, 0.19514355063438416, 0.023306572809815407, 0.3217373192310333], "decide_seconds": 0.1614361319989257, "probabilities_f64_minus_runtime_max_abs": 8.286032443516334e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.3065066038826809, "sagan": 0.06035472883168703, "socrates": 0.34731748404442636, "ibn_sina": 0.06839086761089394, "darwin": 0.21065870272719794, "hypatia": 0.006771612903113822}, "confidence": 0.19376577017418284, "choice": "socrates"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 165.03470400130027, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "c1bb3d6c6a862369b6084f743e6e387dfa5e1afab61d21235ee28083ce7263db", "compiled_ids_sha256": "6d15ea8d0e43ad3952bd2bb8485d0d182cf52eec95bb5533f34210557bdab562", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 16.875, 18.625, 17.0, 18.125, 14.6875], "runtime_probabilities": [0.30650660395622253, 0.060354724526405334, 0.34731748700141907, 0.0683908686041832, 0.2106587141752243, 0.0067716133780777454], "decide_seconds": 0.1615458900014346, "probabilities_f64_minus_runtime_max_abs": 1.1448026360216801e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.1598878245975161, "ibn_sina": 0.031483779386251236, "hypatia": 0.016852052729461577, "einstein": 0.12452076300012897, "sagan": 0.4346201681952752, "darwin": 0.23263541209136682}, "confidence": 0.2009905159435348, "choice": "sagan"}}, "usage": {"input_tokens": 296, "output_tokens": 1}, "latency_ms": 166.71900100118364, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 296, "compiled_prompt_sha256": "4d89b0a4d92b78766d8c6df2302113378d8eabe3158ff0cc3f5de3002ee99f8f", "compiled_ids_sha256": "114a85d8f043627a00e00c3c88c279c6425def97aaa9280ae263fcd587303ed9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 15.75, 15.125, 17.125, 18.375, 17.75], "runtime_probabilities": [0.15988783538341522, 0.03148378059267998, 0.01685205288231373, 0.12452076375484467, 0.4346201717853546, 0.23263542354106903], "decide_seconds": 0.16323229800036643, "probabilities_f64_minus_runtime_max_abs": 1.1449702214116897e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:47:57Z"} {"id": "c-darwin-01-dessert-02", "uid": "c003", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "prompt_sha256_by_order": ["1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2"], "request_sha256": "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "compiled_prompt_sha256_by_order": ["604afd02ea457c03d0d85cec823cbc75b5a82d677e33fef5295394b9bc38e733", "eecb982988a136fdc7afcc5fa85ebd799ece428ecdb90ce6d98a62ec58e25193", "d4814fd3b0c2bfb52fe0d3c351348f1f63c210a78bcb2c3fd48de52aad02360a", "0c93ef4823810b24557e1073fda065276dfb6e7dd48c59c111f2603bc5b15d83", "0e793654ef3c48245547a933abf15a5952a6c5a6caf5cdd53b831304b7eaee5b", "349c3e96f2945599c0a60343dcd53d3a1781af22f0db26192c8b8b3a0b5b1f4d"], "compiled_tokens_by_order": [399, 399, 399, 399, 399, 399], "choice": "darwin", "probs": {"darwin": 0.8609890036474503, "einstein": 0.01083825204708563, "hypatia": 0.010181595551097135, "ibn_sina": 0.006175449866534159, "sagan": 0.10283047235062048, "socrates": 0.008985226537212317}, "confidence": 0.7029328590783799, "seconds": 0.17777628649976882, "seconds_all_orders": 1.0666577189986128, "seconds_by_order": [0.17760134799755178, 0.17769565600247006, 0.17724240100142197, 0.17726774099719478, 0.1770247019994713, 0.17982587100050296], "argmax_stable": true, "spread": {"darwin": 0.32975408472523116, "einstein": 0.033045762597008546, "hypatia": 0.014413859117250082, "ibn_sina": 0.021335947925444514, "sagan": 0.28963624972559376, "socrates": 0.03744575476721382}, "spread_2dp": {"darwin": 0.32999999999999996, "einstein": 0.03, "hypatia": 0.01, "ibn_sina": 0.02, "sagan": 0.29000000000000004, "socrates": 0.03}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.8609890036474503, "einstein": 0.01083825204708563, "hypatia": 0.010181595551097135, "ibn_sina": 0.006175449866534159, "sagan": 0.10283047235062048, "socrates": 0.008985226537212317}, {"ibn_sina": 0.02494045658316166, "hypatia": 0.01178103749997502, "sagan": 0.10500324425514149, "socrates": 0.04377186447299401, "darwin": 0.7758748623709579, "einstein": 0.03862853481776991}, {"hypatia": 0.013061381852331947, "darwin": 0.7591200444648291, "sagan": 0.19193560520976854, "ibn_sina": 0.01677114627545823, "einstein": 0.00955591109880609, "socrates": 0.00955591109880609}, {"ibn_sina": 0.013457298513995743, "einstein": 0.01048055462072554, "darwin": 0.6091238415390786, "sagan": 0.3260404975671823, "hypatia": 0.023618294427093137, "socrates": 0.01727951333192471}, {"einstein": 0.027900738757387456, "sagan": 0.04321350954171735, "socrates": 0.031615678962355245, "ibn_sina": 0.016922653484988052, "darwin": 0.8679665414806864, "hypatia": 0.012380877772865615}, {"socrates": 0.006326109705780192, "ibn_sina": 0.0036045086577171466, "hypatia": 0.009204435309843054, "einstein": 0.0055827722207613655, "sagan": 0.036404247841588515, "darwin": 0.9388779262643098}], "prompt_tokens": 399, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.8609890036474503, "einstein": 0.01083825204708563, "hypatia": 0.010181595551097135, "ibn_sina": 0.006175449866534159, "sagan": 0.10283047235062048, "socrates": 0.008985226537212317}, "confidence": 0.7029328590783799, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.573258002667, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "604afd02ea457c03d0d85cec823cbc75b5a82d677e33fef5295394b9bc38e733", "compiled_ids_sha256": "77f997cd82656949ba572af316c23fd06bc92bfc51d6b9880262378e26c5a2fd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 14.875, 14.8125, 14.3125, 17.125, 14.6875], "runtime_probabilities": [0.8609890937805176, 0.010838253423571587, 0.01018159557133913, 0.006175450049340725, 0.10283048450946808, 0.008985227905213833], "decide_seconds": 0.1727490340017539, "probabilities_f64_minus_runtime_max_abs": 9.01330672320455e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02494045658316166, "hypatia": 0.01178103749997502, "sagan": 0.10500324425514149, "socrates": 0.04377186447299401, "darwin": 0.7758748623709579, "einstein": 0.03862853481776991}, "confidence": 0.5308710999276713, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.6694679974753, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "eecb982988a136fdc7afcc5fa85ebd799ece428ecdb90ce6d98a62ec58e25193", "compiled_ids_sha256": "e07465fed92b998815c200a700fdd85da9cfc4c20951a7dbaa1575588e4be7dc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 14.9375, 17.125, 16.25, 19.125, 16.125], "runtime_probabilities": [0.024940459057688713, 0.011781037785112858, 0.1050032451748848, 0.04377186670899391, 0.7758749127388, 0.038628533482551575], "decide_seconds": 0.17289509400143288, "probabilities_f64_minus_runtime_max_abs": 5.036784211931433e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.013061381852331947, "darwin": 0.7591200444648291, "sagan": 0.19193560520976854, "ibn_sina": 0.01677114627545823, "einstein": 0.00955591109880609, "socrates": 0.00955591109880609}, "confidence": 0.586929549595407, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.2166209993884, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "d4814fd3b0c2bfb52fe0d3c351348f1f63c210a78bcb2c3fd48de52aad02360a", "compiled_ids_sha256": "7f9a490fafd5512098bba5444a69e59e3122a02dc5193c4ac061e78967177a72", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 19.125, 17.75, 15.3125, 14.75, 14.75], "runtime_probabilities": [0.013061381876468658, 0.7591201066970825, 0.19193561375141144, 0.016771147027611732, 0.009555911645293236, 0.009555911645293236], "decide_seconds": 0.1725490360004187, "probabilities_f64_minus_runtime_max_abs": 6.223225346158046e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013457298513995743, "einstein": 0.01048055462072554, "darwin": 0.6091238415390786, "sagan": 0.3260404975671823, "hypatia": 0.023618294427093137, "socrates": 0.01727951333192471}, "confidence": 0.48000279417536196, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.2418810032832, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "0c93ef4823810b24557e1073fda065276dfb6e7dd48c59c111f2603bc5b15d83", "compiled_ids_sha256": "f07dedf64fdfc9daf35a77d0235b5d2fa815c708313d843e26184b6a70649a04", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.0625, 19.125, 18.5, 15.875, 15.5625], "runtime_probabilities": [0.013457298278808594, 0.010480553843080997, 0.6091238260269165, 0.32604050636291504, 0.023618293926119804, 0.01727951318025589], "decide_seconds": 0.17251538700293167, "probabilities_f64_minus_runtime_max_abs": 1.5512162110375982e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.027900738757387456, "sagan": 0.04321350954171735, "socrates": 0.031615678962355245, "ibn_sina": 0.016922653484988052, "darwin": 0.8679665414806864, "hypatia": 0.012380877772865615}, "confidence": 0.6700840401870818, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 176.99977200027206, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "0e793654ef3c48245547a933abf15a5952a6c5a6caf5cdd53b831304b7eaee5b", "compiled_ids_sha256": "ef5fbcd2e719aeca70027a5142e9f5a5f819a3c42c9935b2f27fe7f883518c5c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 16.25, 15.9375, 15.3125, 19.25, 15.0], "runtime_probabilities": [0.02790074050426483, 0.043213509023189545, 0.03161568194627762, 0.01692265458405018, 0.8679665923118591, 0.012380877509713173], "decide_seconds": 0.17244436699911603, "probabilities_f64_minus_runtime_max_abs": 5.0831172715426476e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.006326109705780192, "ibn_sina": 0.0036045086577171466, "hypatia": 0.009204435309843054, "einstein": 0.0055827722207613655, "sagan": 0.036404247841588515, "darwin": 0.9388779262643098}, "confidence": 0.8301968139191056, "choice": "darwin"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 179.8015310014307, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "349c3e96f2945599c0a60343dcd53d3a1781af22f0db26192c8b8b3a0b5b1f4d", "compiled_ids_sha256": "d3752dbfec8ff57afd0889338a3cbf8b0c19159b4effc6f1fc6706057e3787c1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 13.8125, 14.75, 14.25, 16.125, 19.375], "runtime_probabilities": [0.006326110102236271, 0.003604508936405182, 0.009204436093568802, 0.005582772195339203, 0.03640424832701683, 0.9388779401779175], "decide_seconds": 0.17519893600183423, "probabilities_f64_minus_runtime_max_abs": 1.3913607710414055e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:47:58Z"} {"id": "c-darwin-01-dessert-03", "uid": "c004", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "prompt_sha256_by_order": ["7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c"], "request_sha256": "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "compiled_prompt_sha256_by_order": ["f23455f0797804d5728aeff3fdc9d48c4b31d4a7b4b5aa74d195e7140ad387f6", "9fa8721a8c13ea09d37498a6c5184e3a2a581f98cac48c4886410653bc6b6ad0", "7f0f69b8f7631a215fecc71cfdd27c835cfcca1a7c03c3ef269839607f7d3091", "fed544507f2d46f84c7f22c5fec1bbfee70fe8aa08be2e63ec0d02302847bcba", "56018aedd0d7a17a9d82f95e083c24bad6beb6d0a716cf2c5009462115797d77", "cdd784ffeb6b17ba13fea7f684921b658c3adf1c0f1b9ddb2b01afe5a57952ff"], "compiled_tokens_by_order": [409, 409, 409, 409, 409, 409], "choice": "sagan", "probs": {"darwin": 0.06254983152697362, "einstein": 0.0552000325797425, "hypatia": 0.09100946785756152, "ibn_sina": 0.027756319486119268, "sagan": 0.593456278136656, "socrates": 0.1700280704129471}, "confidence": 0.2957701498750316, "seconds": 0.18198334566598837, "seconds_all_orders": 1.0919000739959301, "seconds_by_order": [0.1780336849988089, 0.18500327100264258, 0.18335461300011957, 0.17722696099735913, 0.18240978999892832, 0.18587175399807165], "argmax_stable": false, "spread": {"darwin": 0.23256223078252378, "einstein": 0.2964678471330715, "hypatia": 0.32329939008229347, "ibn_sina": 0.0435887203679296, "sagan": 0.5200164297154015, "socrates": 0.2927147365057329}, "spread_2dp": {"darwin": 0.22999999999999998, "einstein": 0.29000000000000004, "hypatia": 0.32, "ibn_sina": 0.039999999999999994, "sagan": 0.52, "socrates": 0.29000000000000004}, "choices_by_order": ["sagan", "einstein", "sagan", "sagan", "hypatia", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.06254983152697362, "einstein": 0.0552000325797425, "hypatia": 0.09100946785756152, "ibn_sina": 0.027756319486119268, "sagan": 0.593456278136656, "socrates": 0.1700280704129471}, {"ibn_sina": 0.024970080258132226, "hypatia": 0.04965893422782462, "sagan": 0.2521888769486581, "socrates": 0.19640489484951307, "darwin": 0.1529602859078577, "einstein": 0.32381692780801424}, {"hypatia": 0.05335325296863391, "darwin": 0.11294883742093823, "sagan": 0.34790691208967733, "ibn_sina": 0.06045715606749058, "einstein": 0.18622115075675147, "socrates": 0.2391126906965085}, {"ibn_sina": 0.025692083641980824, "einstein": 0.02734908067494276, "darwin": 0.05789800424319521, "sagan": 0.3775424130979813, "hypatia": 0.17833840818861754, "socrates": 0.3331800101532823}, {"einstein": 0.050474420455962295, "sagan": 0.07343984842125445, "socrates": 0.15547216032782799, "ibn_sina": 0.05719501145911825, "darwin": 0.290460235025719, "hypatia": 0.3729583243101181}, {"socrates": 0.040465273647549434, "ibn_sina": 0.016868435699560976, "hypatia": 0.1412376828974904, "einstein": 0.07559908390842286, "sagan": 0.49296795182948433, "darwin": 0.2328615720174921}], "prompt_tokens": 409, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.06254983152697362, "einstein": 0.0552000325797425, "hypatia": 0.09100946785756152, "ibn_sina": 0.027756319486119268, "sagan": 0.593456278136656, "socrates": 0.1700280704129471}, "confidence": 0.2957701498750316, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 178.00418499973603, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "f23455f0797804d5728aeff3fdc9d48c4b31d4a7b4b5aa74d195e7140ad387f6", "compiled_ids_sha256": "2fc9c2e9c165a1094386aba821c96f2b186d61c9c48cddddeb2a7985bc7f809a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 16.625, 17.125, 15.9375, 19.0, 17.75], "runtime_probabilities": [0.06254982948303223, 0.05520002916455269, 0.0910094678401947, 0.02775632031261921, 0.5934562683105469, 0.17002807557582855], "decide_seconds": 0.17318325100131915, "probabilities_f64_minus_runtime_max_abs": 9.826109148569628e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.024970080258132226, "hypatia": 0.04965893422782462, "sagan": 0.2521888769486581, "socrates": 0.19640489484951307, "darwin": 0.1529602859078577, "einstein": 0.32381692780801424}, "confidence": 0.1289884603667616, "choice": "einstein"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 184.97669100179337, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "9fa8721a8c13ea09d37498a6c5184e3a2a581f98cac48c4886410653bc6b6ad0", "compiled_ids_sha256": "d6914ad33a20e73683ad4646cfbf5dd65a305eced249bf7a118714060ab9afc5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 16.625, 18.25, 18.0, 17.75, 18.5], "runtime_probabilities": [0.024970078840851784, 0.04965893551707268, 0.25218886137008667, 0.19640488922595978, 0.15296028554439545, 0.3238169252872467], "decide_seconds": 0.18010881900045206, "probabilities_f64_minus_runtime_max_abs": 1.5578571432861565e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.05335325296863391, "darwin": 0.11294883742093823, "sagan": 0.34790691208967733, "ibn_sina": 0.06045715606749058, "einstein": 0.18622115075675147, "socrates": 0.2391126906965085}, "confidence": 0.10993598798178172, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 183.32858400026453, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "7f0f69b8f7631a215fecc71cfdd27c835cfcca1a7c03c3ef269839607f7d3091", "compiled_ids_sha256": "79ca49438526ba73e13ce1ce1676b217ae575e085426773c613a7e87efcee660", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 17.625, 18.75, 17.0, 18.125, 18.375], "runtime_probabilities": [0.05335325747728348, 0.11294884234666824, 0.3479069173336029, 0.06045715510845184, 0.1862211674451828, 0.2391127049922943], "decide_seconds": 0.17863384999873233, "probabilities_f64_minus_runtime_max_abs": 1.6688431331290943e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.025692083641980824, "einstein": 0.02734908067494276, "darwin": 0.05789800424319521, "sagan": 0.3775424130979813, "hypatia": 0.17833840818861754, "socrates": 0.3331800101532823}, "confidence": 0.21927509524727606, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.20169099993655, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "fed544507f2d46f84c7f22c5fec1bbfee70fe8aa08be2e63ec0d02302847bcba", "compiled_ids_sha256": "e3ce0791ee677404ad2ba5e76cad11df8fcee12777f77958466d2dbbb4a33b6f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 16.0, 16.75, 18.625, 17.875, 18.5], "runtime_probabilities": [0.025692082941532135, 0.0273490808904171, 0.057898007333278656, 0.3775424063205719, 0.1783384084701538, 0.3331800103187561], "decide_seconds": 0.1724743570011924, "probabilities_f64_minus_runtime_max_abs": 6.777409411462543e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.050474420455962295, "sagan": 0.07343984842125445, "socrates": 0.15547216032782799, "ibn_sina": 0.05719501145911825, "darwin": 0.290460235025719, "hypatia": 0.3729583243101181}, "confidence": 0.1502927357575018, "choice": "hypatia"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 182.38464099704288, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "56018aedd0d7a17a9d82f95e083c24bad6beb6d0a716cf2c5009462115797d77", "compiled_ids_sha256": "24863b7310897a5b13d6d6ac85f5b3f83351e7a1c22332e9ba4d2e875f7cb554", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 16.875, 17.625, 16.625, 18.25, 18.5], "runtime_probabilities": [0.050474416464567184, 0.0734398365020752, 0.15547214448451996, 0.05719501152634621, 0.2904602289199829, 0.37295830249786377], "decide_seconds": 0.17775213700224413, "probabilities_f64_minus_runtime_max_abs": 2.1812254336062153e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.040465273647549434, "ibn_sina": 0.016868435699560976, "hypatia": 0.1412376828974904, "einstein": 0.07559908390842286, "sagan": 0.49296795182948433, "darwin": 0.2328615720174921}, "confidence": 0.24189219793161898, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 185.84546399870305, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "cdd784ffeb6b17ba13fea7f684921b658c3adf1c0f1b9ddb2b01afe5a57952ff", "compiled_ids_sha256": "cd690b2a8827e88dec9d609a4e2afab20cdeb3985dc7b8c97f0fc407b0ab1094", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 15.5, 17.625, 17.0, 18.875, 18.125], "runtime_probabilities": [0.04046526923775673, 0.016868432983756065, 0.14123767614364624, 0.07559908181428909, 0.4929679036140442, 0.23286156356334686], "decide_seconds": 0.18091129299864406, "probabilities_f64_minus_runtime_max_abs": 4.821544014044221e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:47:59Z"} {"id": "c-darwin-01-dessert-03", "uid": "c005", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "prompt_sha256_by_order": ["1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06"], "request_sha256": "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "compiled_prompt_sha256_by_order": ["24e4de0b01d855575257e46ab574c911b12ec716af49ef43360d71b0ceb39554", "7ae82cf86dec5f8bd087468bc1c0b3049d4f35086b412a067c128b307da5d00b", "bf6d464337d82fb63d386fdea062a1714e3f546ce4e7485345e36242709deb27", "980e3454da994c6c2b3c6b12697cde2623b178247de8e4b821e04ce52bffdea4", "cb00dc996172d719bcdc935c287e628ad41efa192010278591ebc2679d4083e0", "0d4e9e0115d996b05648a1b53ae79f2fe148aa38b656e90f7f7f749a2c6f3051"], "compiled_tokens_by_order": [411, 411, 411, 411, 411, 411], "choice": "darwin", "probs": {"darwin": 0.9952739540826643, "einstein": 0.001240504349012897, "hypatia": 0.0010947412456466016, "ibn_sina": 0.0008009298049675685, "sagan": 0.0009661057584147277, "socrates": 0.0006237647592939699}, "confidence": 0.9790706867428549, "seconds": 0.17793158016623542, "seconds_all_orders": 1.0675894809974125, "seconds_by_order": [0.17820190299971728, 0.17833275199882337, 0.17777986699729809, 0.1776104979981028, 0.17782122500284459, 0.17784323600062635], "argmax_stable": true, "spread": {"darwin": 0.04765272621080685, "einstein": 0.046151525385624004, "hypatia": 0.0011235338130737486, "ibn_sina": 0.002162289738599398, "sagan": 0.0007881158927879364, "socrates": 0.0012498095308466051}, "spread_2dp": {"darwin": 0.050000000000000044, "einstein": 0.05, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9952739540826643, "einstein": 0.001240504349012897, "hypatia": 0.0010947412456466016, "ibn_sina": 0.0008009298049675685, "sagan": 0.0009661057584147277, "socrates": 0.0006237647592939699}, {"ibn_sina": 0.0029632195435669666, "hypatia": 0.001399725800809531, "sagan": 0.001090107549755712, "socrates": 0.0017972835046330409, "darwin": 0.9910612719994032, "einstein": 0.0016883916018314625}, {"hypatia": 0.0019201037446599225, "darwin": 0.9946383644275474, "sagan": 0.0008004183249912023, "ibn_sina": 0.0010277574732713325, "einstein": 0.0010277574732713325, "socrates": 0.0005855985562588559}, {"ibn_sina": 0.0016863385584010874, "einstein": 0.006265507841229858, "darwin": 0.9898561654166824, "sagan": 0.0008479442783140605, "hypatia": 0.0007965699315861739, "socrates": 0.0005474739737864356}, {"einstein": 0.04717928285889534, "sagan": 0.001516575562418978, "socrates": 0.0014246908940800814, "ibn_sina": 0.0013383732363702386, "darwin": 0.9476212278718574, "hypatia": 0.0009198495763780407}, {"socrates": 0.0015885342177791386, "ibn_sina": 0.0015885342177791386, "hypatia": 0.0008502810946597584, "einstein": 0.0018000450915201518, "sagan": 0.0015885342177791386, "darwin": 0.9925840711604828}], "prompt_tokens": 411, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9952739540826643, "einstein": 0.001240504349012897, "hypatia": 0.0010947412456466016, "ibn_sina": 0.0008009298049675685, "sagan": 0.0009661057584147277, "socrates": 0.0006237647592939699}, "confidence": 0.9790706867428549, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 178.1722930027172, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "24e4de0b01d855575257e46ab574c911b12ec716af49ef43360d71b0ceb39554", "compiled_ids_sha256": "7edad3a35e2849ccf2f8adce479f5cd518ba5078d299c95eded9c3ee2725b892", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 12.6875, 12.5625, 12.25, 12.4375, 12.0], "runtime_probabilities": [0.995274007320404, 0.001240504439920187, 0.0010947412811219692, 0.0008009298471733928, 0.0009661057265475392, 0.0006237647612579167], "decide_seconds": 0.17316267100250116, "probabilities_f64_minus_runtime_max_abs": 5.323773977217172e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0029632195435669666, "hypatia": 0.001399725800809531, "sagan": 0.001090107549755712, "socrates": 0.0017972835046330409, "darwin": 0.9910612719994032, "einstein": 0.0016883916018314625}, "confidence": 0.963765483227447, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 178.30653299824917, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "7ae82cf86dec5f8bd087468bc1c0b3049d4f35086b412a067c128b307da5d00b", "compiled_ids_sha256": "338b01856f416d4a61dc12810e036acb1a8a1cb1ba05c9d1f761350fa5bc444b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 12.8125, 12.5625, 13.0625, 19.375, 13.0], "runtime_probabilities": [0.002963220002129674, 0.0013997259084135294, 0.0010901076020672917, 0.0017972836503759027, 0.9910613894462585, 0.0016883919015526772], "decide_seconds": 0.17327695100175333, "probabilities_f64_minus_runtime_max_abs": 1.1744685535219901e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0019201037446599225, "darwin": 0.9946383644275474, "sagan": 0.0008004183249912023, "ibn_sina": 0.0010277574732713325, "einstein": 0.0010277574732713325, "socrates": 0.0005855985562588559}, "confidence": 0.9768011583427425, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 177.75379700105987, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "bf6d464337d82fb63d386fdea062a1714e3f546ce4e7485345e36242709deb27", "compiled_ids_sha256": "d14b25d3db65b01907cc3f123ee7db381c4af25471f371b300083463f3006978", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.125, 19.375, 12.25, 12.5, 12.5, 11.9375], "runtime_probabilities": [0.00192010379396379, 0.9946383833885193, 0.0008004183182492852, 0.0010277574183419347, 0.0010277574183419347, 0.0005855985800735652], "decide_seconds": 0.17288521400041645, "probabilities_f64_minus_runtime_max_abs": 1.8960971903680957e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0016863385584010874, "einstein": 0.006265507841229858, "darwin": 0.9898561654166824, "sagan": 0.0008479442783140605, "hypatia": 0.0007965699315861739, "socrates": 0.0005474739737864356}, "confidence": 0.9618054773369201, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 177.5851780002995, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "980e3454da994c6c2b3c6b12697cde2623b178247de8e4b821e04ce52bffdea4", "compiled_ids_sha256": "c2279aafb7c24ad5b099f1127ff01aa5cafc0d9df0a980f9fc1cefa9cd26d047", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 14.5625, 19.625, 12.5625, 12.5, 12.125], "runtime_probabilities": [0.0016863386845216155, 0.00626550754532218, 0.9898561239242554, 0.0008479442331008613, 0.0007965699187479913, 0.000547473959159106], "decide_seconds": 0.17271130500012077, "probabilities_f64_minus_runtime_max_abs": 4.149242704976075e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.04717928285889534, "sagan": 0.001516575562418978, "socrates": 0.0014246908940800814, "ibn_sina": 0.0013383732363702386, "darwin": 0.9476212278718574, "hypatia": 0.0009198495763780407}, "confidence": 0.871898971631335, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 177.79431699818815, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "cb00dc996172d719bcdc935c287e628ad41efa192010278591ebc2679d4083e0", "compiled_ids_sha256": "fb3d3717bd9882bff0d11c1fec042a2cfbea9ff0407970b3aea1ec745500632c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 12.8125, 12.75, 12.6875, 19.25, 12.3125], "runtime_probabilities": [0.04717927798628807, 0.0015165755758062005, 0.00142469082493335, 0.0013383731711655855, 0.9476211667060852, 0.0009198494954034686], "decide_seconds": 0.17295993299921975, "probabilities_f64_minus_runtime_max_abs": 6.116577222492481e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0015885342177791386, "ibn_sina": 0.0015885342177791386, "hypatia": 0.0008502810946597584, "einstein": 0.0018000450915201518, "sagan": 0.0015885342177791386, "darwin": 0.9925840711604828}, "confidence": 0.9690304271000849, "choice": "darwin"}}, "usage": {"input_tokens": 411, "output_tokens": 1}, "latency_ms": 177.81723599910038, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 411, "compiled_prompt_sha256": "0d4e9e0115d996b05648a1b53ae79f2fe148aa38b656e90f7f7f749a2c6f3051", "compiled_ids_sha256": "4ecfe7f6c8cad67259c06bd9e8a387fa43c38e9c23586ef1d7e4501f297ed7ef", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.9375, 12.9375, 12.3125, 13.0625, 12.9375, 19.375], "runtime_probabilities": [0.0015885343309491873, 0.0015885343309491873, 0.0008502810960635543, 0.0018000450218096375, 0.0015885343309491873, 0.9925841093063354], "decide_seconds": 0.17295547400135547, "probabilities_f64_minus_runtime_max_abs": 3.8145852698789895e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:00Z"} {"id": "c-darwin-01-main-01", "uid": "c006", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "prompt_sha256_by_order": ["56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545"], "request_sha256": "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "compiled_prompt_sha256_by_order": ["377f0fd9228b401ffd1752b4418d7a4f949f5c9987b66900bf4451b634fa3b16", "8e484681f634b2f4c1cd7123121993b1b27afe0c195f941c7734f878d0bbd2fd", "7878a73c4f5ae760d37908f3a6ceb98ac84c9a1a487cce85de596986d9a825d8", "9c31b0fac9c51e308e2ab23a9c928df2a8a5f926e559d266b4f0aae4a6dd586b", "727c8ae7ff231dce223ce04956fa1362ac85079812affa37eae8833c4e87dec2", "7ab880c611545781cc4584e295dca04b32e54cbfd963b82802eff6a3c2ca1ddd"], "compiled_tokens_by_order": [314, 314, 314, 314, 314, 314], "choice": "darwin", "probs": {"darwin": 0.9006743260341009, "einstein": 0.01000558796999655, "hypatia": 0.012069042872652501, "ibn_sina": 0.02400215778283913, "sagan": 0.039572868079267506, "socrates": 0.013676017261143389}, "confidence": 0.7378968531392719, "seconds": 0.1659611988331259, "seconds_all_orders": 0.9957671929987555, "seconds_by_order": [0.1656207900014124, 0.1658140280014777, 0.1654019720008364, 0.16516920299909543, 0.16559842999777175, 0.16816276999816182], "argmax_stable": true, "spread": {"darwin": 0.6099412693875745, "einstein": 0.04902325858149747, "hypatia": 0.030061339634121144, "ibn_sina": 0.2662840438703018, "sagan": 0.17662098558446906, "socrates": 0.09412011521116904}, "spread_2dp": {"darwin": 0.61, "einstein": 0.04, "hypatia": 0.03, "ibn_sina": 0.26, "sagan": 0.18, "socrates": 0.09000000000000001}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9006743260341009, "einstein": 0.01000558796999655, "hypatia": 0.012069042872652501, "ibn_sina": 0.02400215778283913, "sagan": 0.039572868079267506, "socrates": 0.013676017261143389}, {"ibn_sina": 0.08441571747755525, "hypatia": 0.007376163188776507, "sagan": 0.031054806971729382, "socrates": 0.04518447752649172, "darwin": 0.8009140278637177, "einstein": 0.031054806971729382}, {"hypatia": 0.028943863459706468, "darwin": 0.3526084417973, "sagan": 0.18873769826427803, "ibn_sina": 0.27461173058932514, "einstein": 0.05407425590106683, "socrates": 0.10102400998832359}, {"ibn_sina": 0.03978749030263556, "einstein": 0.011399306826733381, "darwin": 0.7052501403427573, "sagan": 0.15736257675938206, "hypatia": 0.0351123369536905, "socrates": 0.05108814881480109}, {"einstein": 0.019031931030851924, "sagan": 0.01577802499977737, "socrates": 0.03340207918664389, "ibn_sina": 0.06240329941027842, "darwin": 0.8614509761645455, "hypatia": 0.007933689207903011}, {"socrates": 0.006903894777154543, "ibn_sina": 0.008327686719023331, "hypatia": 0.005050997319569357, "einstein": 0.005050997319569357, "sagan": 0.012116712679808964, "darwin": 0.9625497111848745}], "prompt_tokens": 314, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9006743260341009, "einstein": 0.01000558796999655, "hypatia": 0.012069042872652501, "ibn_sina": 0.02400215778283913, "sagan": 0.039572868079267506, "socrates": 0.013676017261143389}, "confidence": 0.7378968531392719, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 165.59043999950518, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "377f0fd9228b401ffd1752b4418d7a4f949f5c9987b66900bf4451b634fa3b16", "compiled_ids_sha256": "01e1272f80d0db6c2b279d6449fa670109af77f7a80b594d0041b4894a23f3be", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 15.0, 15.1875, 15.875, 16.375, 15.3125], "runtime_probabilities": [0.9006743431091309, 0.01000558864325285, 0.012069043703377247, 0.024002159014344215, 0.03957286849617958, 0.01367601752281189], "decide_seconds": 0.1617373189983482, "probabilities_f64_minus_runtime_max_abs": 1.7075029945523568e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.08441571747755525, "hypatia": 0.007376163188776507, "sagan": 0.031054806971729382, "socrates": 0.04518447752649172, "darwin": 0.8009140278637177, "einstein": 0.031054806971729382}, "confidence": 0.5656367779695031, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 165.78851799931726, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "8e484681f634b2f4c1cd7123121993b1b27afe0c195f941c7734f878d0bbd2fd", "compiled_ids_sha256": "bb9f7932058a5211dd00b00544533586a3bca0139e18ab3f9af64d9bcb455221", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 14.6875, 16.125, 16.5, 19.375, 16.125], "runtime_probabilities": [0.08441571146249771, 0.007376162335276604, 0.031054804101586342, 0.04518447816371918, 0.8009139895439148, 0.031054804101586342], "decide_seconds": 0.16199036699981662, "probabilities_f64_minus_runtime_max_abs": 3.8319802886377374e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.028943863459706468, "darwin": 0.3526084417973, "sagan": 0.18873769826427803, "ibn_sina": 0.27461173058932514, "einstein": 0.05407425590106683, "socrates": 0.10102400998832359}, "confidence": 0.1466265483691317, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 165.37560199867585, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "7878a73c4f5ae760d37908f3a6ceb98ac84c9a1a487cce85de596986d9a825d8", "compiled_ids_sha256": "a1267f356df856ccb6a9b8ce3f42b133d01c199f5bd9d78edee44e5a39f9f0be", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 18.625, 18.0, 18.375, 16.75, 17.375], "runtime_probabilities": [0.0289438646286726, 0.35260844230651855, 0.18873772025108337, 0.2746117413043976, 0.05407426133751869, 0.10102401673793793], "decide_seconds": 0.1614565509989916, "probabilities_f64_minus_runtime_max_abs": 2.1986805348594984e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03978749030263556, "einstein": 0.011399306826733381, "darwin": 0.7052501403427573, "sagan": 0.15736257675938206, "hypatia": 0.0351123369536905, "socrates": 0.05108814881480109}, "confidence": 0.4496464570397559, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 165.14386400012881, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "9c31b0fac9c51e308e2ab23a9c928df2a8a5f926e559d266b4f0aae4a6dd586b", "compiled_ids_sha256": "6db79f55d0ee6ea016effba61e5d8207f47d881faf68f4a0af77584c3c5a6554", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 15.375, 19.5, 18.0, 16.5, 16.875], "runtime_probabilities": [0.03978749364614487, 0.011399307288229465, 0.7052501440048218, 0.15736258029937744, 0.03511233627796173, 0.051088154315948486], "decide_seconds": 0.1613945219978632, "probabilities_f64_minus_runtime_max_abs": 5.501147398723205e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.019031931030851924, "sagan": 0.01577802499977737, "socrates": 0.03340207918664389, "ibn_sina": 0.06240329941027842, "darwin": 0.8614509761645455, "hypatia": 0.007933689207903011}, "confidence": 0.6682796638876979, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 165.5742900002224, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "727c8ae7ff231dce223ce04956fa1362ac85079812affa37eae8833c4e87dec2", "compiled_ids_sha256": "f3fb0b57ba2996e7cb8625328b93d66b8b4602ed8f5d038b5404cc574cb40439", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 15.5, 16.25, 16.875, 19.5, 14.8125], "runtime_probabilities": [0.019031930714845657, 0.01577802561223507, 0.03340207785367966, 0.062403302639722824, 0.8614509701728821, 0.007933689281344414], "decide_seconds": 0.1618049680000695, "probabilities_f64_minus_runtime_max_abs": 5.991663387661106e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.006903894777154543, "ibn_sina": 0.008327686719023331, "hypatia": 0.005050997319569357, "einstein": 0.005050997319569357, "sagan": 0.012116712679808964, "darwin": 0.9625497111848745}, "confidence": 0.8784099360888611, "choice": "darwin"}}, "usage": {"input_tokens": 314, "output_tokens": 1}, "latency_ms": 168.13713000010466, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 314, "compiled_prompt_sha256": "7ab880c611545781cc4584e295dca04b32e54cbfd963b82802eff6a3c2ca1ddd", "compiled_ids_sha256": "8b611d8553e8f365c2aad1be5e33903179b0ffdd2caa1f5940bf817d74748658", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.75, 14.25, 14.25, 15.125, 19.5], "runtime_probabilities": [0.006903894245624542, 0.008327686227858067, 0.005050997249782085, 0.005050997249782085, 0.012116712518036366, 0.9625496864318848], "decide_seconds": 0.16426068999862764, "probabilities_f64_minus_runtime_max_abs": 2.475298976900575e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:01Z"} {"id": "c-darwin-01-main-02", "uid": "c007", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "prompt_sha256_by_order": ["7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf"], "request_sha256": "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "compiled_prompt_sha256_by_order": ["98b6551eb87406f35159db42e99ab68b6616d9349e25272937a35b9bb812bc44", "a5cdaad818d34b9a9d13e3f249d193a4388ffe12a104296b34eeedb316bd586d", "e8618ebc3c93376c0d6c137dc53cca0882fcc0f2c62737655841c3f0f9f0e650", "8d3c1b75fc680d29a518eb5fa9bbda1574c0e9b4b3b1b2250a450711dc27cc9f", "d2bea664d466f575873c7433bfa2ddd0d814a958f44532e762cb9be6a106ac51", "612b973b2241207c94ab421257e98cc6d1faba4ecac98ae2e27795cd7dcb2c6d"], "compiled_tokens_by_order": [295, 295, 295, 295, 295, 295], "choice": "socrates", "probs": {"darwin": 0.09865106900495074, "einstein": 0.020678387418066397, "hypatia": 0.024942896412917904, "ibn_sina": 0.023431682714698462, "sagan": 0.39017254671271234, "socrates": 0.44212341773665403}, "confidence": 0.3209002884565396, "seconds": 0.16655435750059647, "seconds_all_orders": 0.9993261450035789, "seconds_by_order": [0.1652793120010756, 0.1656257499998901, 0.1650553139988915, 0.16508684400105267, 0.17161852400022326, 0.16666040100244572], "argmax_stable": false, "spread": {"darwin": 0.5708423393199984, "einstein": 0.15636084957115853, "hypatia": 0.020357576836836797, "ibn_sina": 0.030879858832882398, "sagan": 0.3479275114691323, "socrates": 0.529837676151459}, "spread_2dp": {"darwin": 0.5700000000000001, "einstein": 0.15, "hypatia": 0.019999999999999997, "ibn_sina": 0.03, "sagan": 0.35000000000000003, "socrates": 0.53}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "darwin", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.09865106900495074, "einstein": 0.020678387418066397, "hypatia": 0.024942896412917904, "ibn_sina": 0.023431682714698462, "sagan": 0.39017254671271234, "socrates": 0.44212341773665403}, {"ibn_sina": 0.03217759532763622, "hypatia": 0.013413617376676169, "sagan": 0.0991139712858512, "socrates": 0.3459417516636459, "darwin": 0.3459417516636459, "einstein": 0.1634113126825446}, {"hypatia": 0.029212617274490628, "darwin": 0.08998119592987996, "sagan": 0.13092186755505358, "ibn_sina": 0.04250410733291122, "einstein": 0.04250410733291122, "socrates": 0.6648761045747533}, {"ibn_sina": 0.019864982758761796, "einstein": 0.01131872598951838, "darwin": 0.15625019261407216, "sagan": 0.09477053240643925, "hypatia": 0.01753078575450368, "socrates": 0.7002647804767047}, {"einstein": 0.06146616358965232, "sagan": 0.04224503524358005, "socrates": 0.18932911272719907, "ibn_sina": 0.037281112752036466, "darwin": 0.6608235352498784, "hypatia": 0.00885504043765383}, {"socrates": 0.7191667888786581, "ibn_sina": 0.01162424850002882, "hypatia": 0.010258363296149065, "einstein": 0.00705046311138607, "sagan": 0.11028779906446029, "darwin": 0.14161233714931754}], "prompt_tokens": 295, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.09865106900495074, "einstein": 0.020678387418066397, "hypatia": 0.024942896412917904, "ibn_sina": 0.023431682714698462, "sagan": 0.39017254671271234, "socrates": 0.44212341773665403}, "confidence": 0.3209002884565396, "choice": "socrates"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 165.25386199646164, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "98b6551eb87406f35159db42e99ab68b6616d9349e25272937a35b9bb812bc44", "compiled_ids_sha256": "1e5073f8ebc6dc8593ad0d9a35127317ad2b076509d4a0a08bc1f851b7949344", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 15.6875, 15.875, 15.8125, 18.625, 18.75], "runtime_probabilities": [0.09865108132362366, 0.020678389817476273, 0.024942899122834206, 0.0234316848218441, 0.3901725709438324, 0.4421234428882599], "decide_seconds": 0.16155380100099137, "probabilities_f64_minus_runtime_max_abs": 2.5151605853590553e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03217759532763622, "hypatia": 0.013413617376676169, "sagan": 0.0991139712858512, "socrates": 0.3459417516636459, "darwin": 0.3459417516636459, "einstein": 0.1634113126825446}, "confidence": 0.20304489279401683, "choice": "socrates"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 165.60344999743393, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "a5cdaad818d34b9a9d13e3f249d193a4388ffe12a104296b34eeedb316bd586d", "compiled_ids_sha256": "a3ae59e3cdd049d8dd5a4241c2590bc4e237f2e3e70b0bf3f52141729134efee", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 15.375, 17.375, 18.625, 18.625, 17.875], "runtime_probabilities": [0.03217759355902672, 0.013413617387413979, 0.09911397099494934, 0.3459417521953583, 0.3459417521953583, 0.16341131925582886], "decide_seconds": 0.16205159699893557, "probabilities_f64_minus_runtime_max_abs": 6.573284250244171e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.029212617274490628, "darwin": 0.08998119592987996, "sagan": 0.13092186755505358, "ibn_sina": 0.04250410733291122, "einstein": 0.04250410733291122, "socrates": 0.6648761045747533}, "confidence": 0.371608170693977, "choice": "socrates"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 165.03275400100392, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "e8618ebc3c93376c0d6c137dc53cca0882fcc0f2c62737655841c3f0f9f0e650", "compiled_ids_sha256": "910f450841c9c0d8ba700d9d2116bedbccf5283d2c303b051072c23db9aec7fe", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 17.5, 17.875, 16.75, 16.75, 19.5], "runtime_probabilities": [0.02921261638402939, 0.08998119086027145, 0.1309218555688858, 0.04250410571694374, 0.04250410571694374, 0.6648761034011841], "decide_seconds": 0.1613134330000321, "probabilities_f64_minus_runtime_max_abs": 1.1986167780531076e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.019864982758761796, "einstein": 0.01131872598951838, "darwin": 0.15625019261407216, "sagan": 0.09477053240643925, "hypatia": 0.01753078575450368, "socrates": 0.7002647804767047}, "confidence": 0.4629208628993652, "choice": "socrates"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 165.06447399660829, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "8d3c1b75fc680d29a518eb5fa9bbda1574c0e9b4b3b1b2250a450711dc27cc9f", "compiled_ids_sha256": "0a673aa79942148ced5ad64143b19878f6fc0f3e35a82c3eaf5fd01c73e6d1fc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.25, 17.875, 17.375, 15.6875, 19.375], "runtime_probabilities": [0.01986498199403286, 0.011318725533783436, 0.15625019371509552, 0.09477052837610245, 0.017530785873532295, 0.7002647519111633], "decide_seconds": 0.1615345910031465, "probabilities_f64_minus_runtime_max_abs": 2.8565541376757153e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.06146616358965232, "sagan": 0.04224503524358005, "socrates": 0.18932911272719907, "ibn_sina": 0.037281112752036466, "darwin": 0.6608235352498784, "hypatia": 0.00885504043765383}, "confidence": 0.4092642474097583, "choice": "darwin"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 171.59569399882457, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "d2bea664d466f575873c7433bfa2ddd0d814a958f44532e762cb9be6a106ac51", "compiled_ids_sha256": "bc3c3b2e426a7c32f6dbef6abe826301cdfffb463cbcef5e3ce2be6fcadce610", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 16.5, 18.0, 16.375, 19.25, 14.9375], "runtime_probabilities": [0.06146616488695145, 0.042245034128427505, 0.189329132437706, 0.03728111833333969, 0.6608235836029053, 0.008855041116476059], "decide_seconds": 0.16799654199712677, "probabilities_f64_minus_runtime_max_abs": 4.835302691574839e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.7191667888786581, "ibn_sina": 0.01162424850002882, "hypatia": 0.010258363296149065, "einstein": 0.00705046311138607, "sagan": 0.11028779906446029, "darwin": 0.14161233714931754}, "confidence": 0.5028751782091676, "choice": "socrates"}}, "usage": {"input_tokens": 295, "output_tokens": 1}, "latency_ms": 166.63620100007392, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 295, "compiled_prompt_sha256": "612b973b2241207c94ab421257e98cc6d1faba4ecac98ae2e27795cd7dcb2c6d", "compiled_ids_sha256": "27119e9fb6228f170ae6889611be15eac66a0a3b73c27ea4e28fbd8e15a56251", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 15.375, 15.25, 14.875, 17.625, 17.875], "runtime_probabilities": [0.7191667556762695, 0.011624248698353767, 0.010258362628519535, 0.00705046346411109, 0.11028780043125153, 0.14161232113838196], "decide_seconds": 0.16272522200233652, "probabilities_f64_minus_runtime_max_abs": 3.3202388616260237e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:02Z"} {"id": "c-darwin-01-main-02", "uid": "c008", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "prompt_sha256_by_order": ["99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce"], "request_sha256": "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "compiled_prompt_sha256_by_order": ["a94c81da04c23301082578b21ecddd1e9b9cf59dce663b78a8e4f45435535a31", "116fca7c5b486793578615fa23511bb9dfc0faca55af0b49790f42fdf6a70204", "ce5c61932522695a78d607f502da7286e6efaeb83ce3fee25065467dd752e5e5", "4891f1a20d154ecbf966cb82a1e1f02fde085fb655fafbc89560787dec24a9b0", "a93b82d3dcdd041fd213f1424827f4178fade3c3c3a8c7f316740083c0934f77", "ec6e43a445db0ba3c7b900c68a8a7ba2f5cdc80561b30f7839eec9c7e6dd58d6"], "compiled_tokens_by_order": [372, 372, 372, 372, 372, 372], "choice": "darwin", "probs": {"darwin": 0.9956817474212499, "einstein": 0.0008529346689396955, "hypatia": 0.0009665016006760816, "ibn_sina": 0.0005506960121401764, "sagan": 0.0012410126205374748, "socrates": 0.0007071076764565674}, "confidence": 0.9806823503078224, "seconds": 0.1719372546673791, "seconds_all_orders": 1.0316235280042747, "seconds_by_order": [0.17247233700254583, 0.17244369799664128, 0.1715609040002164, 0.17109257800257183, 0.17291826400105492, 0.17113574700124445], "argmax_stable": true, "spread": {"darwin": 0.005846878065501326, "einstein": 0.001600260268926288, "hypatia": 0.001606370740666515, "ibn_sina": 0.001902859324815646, "sagan": 0.003461743411272732, "socrates": 0.001326961140946733}, "spread_2dp": {"darwin": 0.010000000000000009, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9956817474212499, "einstein": 0.0008529346689396955, "hypatia": 0.0009665016006760816, "ibn_sina": 0.0005506960121401764, "sagan": 0.0012410126205374748, "socrates": 0.0007071076764565674}, {"ibn_sina": 0.0024535553369558222, "hypatia": 0.001686302278010108, "sagan": 0.001686302278010108, "socrates": 0.0020340688174033004, "darwin": 0.9898348693557486, "einstein": 0.0023049019338720186}, {"hypatia": 0.0015929490200378145, "darwin": 0.9953426282946258, "sagan": 0.0008526441680233662, "ibn_sina": 0.00075245583728744, "einstein": 0.0007068668427380724, "socrates": 0.00075245583728744}, {"ibn_sina": 0.0014029053596561105, "einstein": 0.00102638748705742, "darwin": 0.9933125255185321, "sagan": 0.0016922268818920208, "hypatia": 0.0014029053596561105, "socrates": 0.0011630493932062627}, {"einstein": 0.0023071271116643604, "sagan": 0.0014895932199279328, "socrates": 0.0017967924012094725, "ibn_sina": 0.0011600963661377297, "darwin": 0.9907904668746745, "hypatia": 0.0024559240263858774}, {"socrates": 0.0011612016639700872, "ibn_sina": 0.001090848011694244, "hypatia": 0.0008495532857193623, "einstein": 0.0008495532857193623, "sagan": 0.004314387579296098, "darwin": 0.9917344561736008}], "prompt_tokens": 372, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9956817474212499, "einstein": 0.0008529346689396955, "hypatia": 0.0009665016006760816, "ibn_sina": 0.0005506960121401764, "sagan": 0.0012410126205374748, "socrates": 0.0007071076764565674}, "confidence": 0.9806823503078224, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 172.44133699932718, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "a94c81da04c23301082578b21ecddd1e9b9cf59dce663b78a8e4f45435535a31", "compiled_ids_sha256": "6aa68021cc66abd7749c997fa9d95159d904c84b3398aad7280decc31822aa57", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 12.6875, 12.8125, 12.25, 13.0625, 12.5], "runtime_probabilities": [0.995681643486023, 0.0008529346087016165, 0.0009665014804340899, 0.0005506959278136492, 0.0012410125927999616, 0.0007071075960993767], "decide_seconds": 0.16793126199991093, "probabilities_f64_minus_runtime_max_abs": 1.039352269804894e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0024535553369558222, "hypatia": 0.001686302278010108, "sagan": 0.001686302278010108, "socrates": 0.0020340688174033004, "darwin": 0.9898348693557486, "einstein": 0.0023049019338720186}, "confidence": 0.9592589828610797, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 172.41696799828787, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "116fca7c5b486793578615fa23511bb9dfc0faca55af0b49790f42fdf6a70204", "compiled_ids_sha256": "0638fc3f06490b8f739806553d9dd009120b9c42242235454840e71c731ba0d5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 13.375, 13.375, 13.5625, 19.75, 13.6875], "runtime_probabilities": [0.002453555352985859, 0.0016863024793565273, 0.0016863024793565273, 0.0020340688060969114, 0.9898348450660706, 0.002304901834577322], "decide_seconds": 0.16785290200277814, "probabilities_f64_minus_runtime_max_abs": 2.4289678046685026e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0015929490200378145, "darwin": 0.9953426282946258, "sagan": 0.0008526441680233662, "ibn_sina": 0.00075245583728744, "einstein": 0.0007068668427380724, "socrates": 0.00075245583728744}, "confidence": 0.9794135330600643, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 171.53481399873272, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "ce5c61932522695a78d607f502da7286e6efaeb83ce3fee25065467dd752e5e5", "compiled_ids_sha256": "29a415f9cc2e86b9c574bc0d3b6b13218727624569c4575ca1e736cff56928a4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 19.625, 12.5625, 12.4375, 12.375, 12.4375], "runtime_probabilities": [0.0015929490327835083, 0.9953426122665405, 0.0008526441524736583, 0.0007524558459408581, 0.0007068667910061777, 0.0007524558459408581], "decide_seconds": 0.16724117700141505, "probabilities_f64_minus_runtime_max_abs": 1.602808530343225e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0014029053596561105, "einstein": 0.00102638748705742, "darwin": 0.9933125255185321, "sagan": 0.0016922268818920208, "hypatia": 0.0014029053596561105, "socrates": 0.0011630493932062627}, "confidence": 0.9716379272426188, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 171.06470799990348, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "4891f1a20d154ecbf966cb82a1e1f02fde085fb655fafbc89560787dec24a9b0", "compiled_ids_sha256": "263af896e08189ae7ec587ee49884bd10bbb6f703096e4710444629843c716df", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 12.875, 19.75, 13.375, 13.1875, 13.0], "runtime_probabilities": [0.0014029054436832666, 0.0010263875592499971, 0.9933125376701355, 0.0016922270879149437, 0.0014029054436832666, 0.0011630493681877851], "decide_seconds": 0.16666701099893544, "probabilities_f64_minus_runtime_max_abs": 1.2151603390186949e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0023071271116643604, "sagan": 0.0014895932199279328, "socrates": 0.0017967924012094725, "ibn_sina": 0.0011600963661377297, "darwin": 0.9907904668746745, "hypatia": 0.0024559240263858774}, "confidence": 0.9627014870512479, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 172.89260400139028, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "a93b82d3dcdd041fd213f1424827f4178fade3c3c3a8c7f316740083c0934f77", "compiled_ids_sha256": "86716e25889c79faa630702c84b7931c01c704a2c123e2e1593818c54db5555e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.4375, 13.0, 13.1875, 12.75, 19.5, 13.5], "runtime_probabilities": [0.002307126997038722, 0.001489593181759119, 0.001796792377717793, 0.0011600963771343231, 0.9907904863357544, 0.0024559241719543934], "decide_seconds": 0.168614575999527, "probabilities_f64_minus_runtime_max_abs": 1.9461079858729136e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0011612016639700872, "ibn_sina": 0.001090848011694244, "hypatia": 0.0008495532857193623, "einstein": 0.0008495532857193623, "sagan": 0.004314387579296098, "darwin": 0.9917344561736008}, "confidence": 0.9670553833878457, "choice": "darwin"}}, "usage": {"input_tokens": 372, "output_tokens": 1}, "latency_ms": 171.11046799982432, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 372, "compiled_prompt_sha256": "ec6e43a445db0ba3c7b900c68a8a7ba2f5cdc80561b30f7839eec9c7e6dd58d6", "compiled_ids_sha256": "7b6a22dceaeea48800c4fd7032f2cb4080502ef42f714548914c636bf2e5fed2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.75, 12.6875, 12.4375, 12.4375, 14.0625, 19.5], "runtime_probabilities": [0.001161201624199748, 0.001090848003514111, 0.0008495533256791532, 0.0008495533256791532, 0.004314387682825327, 0.991734504699707], "decide_seconds": 0.166740390999621, "probabilities_f64_minus_runtime_max_abs": 4.852610624439535e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:03Z"} {"id": "c-darwin-01-main-02", "uid": "c009", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "prompt_sha256_by_order": ["e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1"], "request_sha256": "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "compiled_prompt_sha256_by_order": ["67d1a2d7e7aebb688c009d8c350b3caab3a77e09611db29376cf5a84ebb7f2f5", "6b489c00d800c69dea62a687b738e00186272d8e8645d2b0ffc3bf9e329b17f2", "3152d2d4b8bd9212d970810c5ccb53ed516fad4fa875eec7f36807d575252e9e", "a1d8850e960e60d0520aecdda88b8f09a7c59bf25dd5970dbaf74baa3604f978", "43530d513862b058270391af2e61b92fc71bee46b380f65227e9983640227649", "10e933efb739aa29898f0d6ec85216a1e3a887f656ea440e3dfb73055489427a"], "compiled_tokens_by_order": [407, 407, 407, 407, 407, 407], "choice": "darwin", "probs": {"darwin": 0.9933585419414035, "einstein": 0.001402970350970143, "hypatia": 0.001317968674441359, "ibn_sina": 0.0018014495895049846, "sagan": 0.001026435035718512, "socrates": 0.001092634407961511}, "confidence": 0.9718298631868472, "seconds": 0.17757412599955083, "seconds_all_orders": 1.065444755997305, "seconds_by_order": [0.17755823799961945, 0.17742803899818682, 0.17786865499874693, 0.17756218800059287, 0.17758594699989771, 0.17744168900026125], "argmax_stable": true, "spread": {"darwin": 0.005274417810687382, "einstein": 0.0015576306488845467, "hypatia": 0.0009461689224727537, "ibn_sina": 0.0035576132631476172, "sagan": 0.0005091370046815423, "socrates": 0.0009068542140939198}, "spread_2dp": {"darwin": 0.010000000000000009, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9933585419414035, "einstein": 0.001402970350970143, "hypatia": 0.001317968674441359, "ibn_sina": 0.0018014495895049846, "sagan": 0.001026435035718512, "socrates": 0.001092634407961511}, {"ibn_sina": 0.004586383240411982, "hypatia": 0.001159619284711429, "sagan": 0.0007969939018614211, "socrates": 0.0014889806352507333, "darwin": 0.9903830113021042, "einstein": 0.0015850116356604}, {"hypatia": 0.0014968514845366133, "darwin": 0.9956182408495604, "sagan": 0.0005172980310369697, "ibn_sina": 0.001028769977264365, "einstein": 0.0007526641939576908, "socrates": 0.0005861754636440574}, {"ibn_sina": 0.0014969104017074598, "einstein": 0.0009664779950570054, "darwin": 0.9956574291127915, "sagan": 0.000664249964190074, "hypatia": 0.0005506825620638596, "socrates": 0.000664249964190074}, {"einstein": 0.0023102948428422375, "sagan": 0.001025187122005434, "socrates": 0.0009630741742400572, "ibn_sina": 0.0024592960586720686, "darwin": 0.9921508417915114, "hypatia": 0.0010913060107287321}, {"socrates": 0.0014930296777379772, "ibn_sina": 0.0029692440633158896, "hypatia": 0.0007052572820131164, "einstein": 0.0008507026671886443, "sagan": 0.0009055682754089559, "darwin": 0.9930761980343356}], "prompt_tokens": 407, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9933585419414035, "einstein": 0.001402970350970143, "hypatia": 0.001317968674441359, "ibn_sina": 0.0018014495895049846, "sagan": 0.001026435035718512, "socrates": 0.001092634407961511}, "confidence": 0.9718298631868472, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.5293380014773, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "67d1a2d7e7aebb688c009d8c350b3caab3a77e09611db29376cf5a84ebb7f2f5", "compiled_ids_sha256": "b6064067571c6aaaef5c497a87a07eae439d78eab1c8465b4f06fd5fd0319cf5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 12.6875, 12.625, 12.9375, 12.375, 12.4375], "runtime_probabilities": [0.9933585524559021, 0.001402970403432846, 0.0013179687084630132, 0.0018014495726674795, 0.0010264350567013025, 0.0010926343966275454], "decide_seconds": 0.17272453499754192, "probabilities_f64_minus_runtime_max_abs": 1.051449860511866e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004586383240411982, "hypatia": 0.001159619284711429, "sagan": 0.0007969939018614211, "socrates": 0.0014889806352507333, "darwin": 0.9903830113021042, "einstein": 0.0015850116356604}, "confidence": 0.9622141030057914, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.4024689984799, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "6b489c00d800c69dea62a687b738e00186272d8e8645d2b0ffc3bf9e329b17f2", "compiled_ids_sha256": "6cf73cd9168f1c04c249ee559e887ecf70908f26037ca053437bb6e75d74ce6a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.875, 12.5, 12.125, 12.75, 19.25, 12.8125], "runtime_probabilities": [0.004586383234709501, 0.0011596191907301545, 0.0007969938451424241, 0.0014889804879203439, 0.990382969379425, 0.0015850116033107042], "decide_seconds": 0.1726961249987653, "probabilities_f64_minus_runtime_max_abs": 4.19226791059657e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0014968514845366133, "darwin": 0.9956182408495604, "sagan": 0.0005172980310369697, "ibn_sina": 0.001028769977264365, "einstein": 0.0007526641939576908, "socrates": 0.0005861754636440574}, "confidence": 0.9805357603521523, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.84254500293173, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "3152d2d4b8bd9212d970810c5ccb53ed516fad4fa875eec7f36807d575252e9e", "compiled_ids_sha256": "8ad081105ad1e05226a47cacb407c9b3202a64ae11e9af9a01fb847c4e7ea348", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.875, 19.375, 11.8125, 12.5, 12.1875, 11.9375], "runtime_probabilities": [0.001496851327829063, 0.9956182241439819, 0.0005172980017960072, 0.0010287698823958635, 0.0007526641711592674, 0.0005861754179932177], "decide_seconds": 0.17312331199718756, "probabilities_f64_minus_runtime_max_abs": 1.6705578476106098e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0014969104017074598, "einstein": 0.0009664779950570054, "darwin": 0.9956574291127915, "sagan": 0.000664249964190074, "hypatia": 0.0005506825620638596, "socrates": 0.000664249964190074}, "confidence": 0.9806717020487398, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.53577799885534, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "a1d8850e960e60d0520aecdda88b8f09a7c59bf25dd5970dbaf74baa3604f978", "compiled_ids_sha256": "9e5b8395a56beaa0a1896dfc8384c81c6512380e4eb2cdf6f58faa5b4b8e1561", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.875, 12.4375, 19.375, 12.0625, 11.875, 12.0625], "runtime_probabilities": [0.0014969103503972292, 0.0009664779645390809, 0.9956574440002441, 0.0006642499938607216, 0.0005506825400516391, 0.0006642499938607216], "decide_seconds": 0.17267150599946035, "probabilities_f64_minus_runtime_max_abs": 1.4887452604028795e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0023102948428422375, "sagan": 0.001025187122005434, "socrates": 0.0009630741742400572, "ibn_sina": 0.0024592960586720686, "darwin": 0.9921508417915114, "hypatia": 0.0010913060107287321}, "confidence": 0.9677378154966789, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.55931699866778, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "43530d513862b058270391af2e61b92fc71bee46b380f65227e9983640227649", "compiled_ids_sha256": "c7bb9afc45b9b132c444c5d82278775f4ca5ed974d8270acb37de6e52bb95fa4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 12.5, 12.4375, 13.375, 19.375, 12.5625], "runtime_probabilities": [0.0023102948907762766, 0.0010251872008666396, 0.000963074155151844, 0.0024592962581664324, 0.9921509027481079, 0.0010913059813901782], "decide_seconds": 0.1726874650012178, "probabilities_f64_minus_runtime_max_abs": 6.095659654814511e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0014930296777379772, "ibn_sina": 0.0029692440633158896, "hypatia": 0.0007052572820131164, "einstein": 0.0008507026671886443, "sagan": 0.0009055682754089559, "darwin": 0.9930761980343356}, "confidence": 0.9713290269060006, "choice": "darwin"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.41490900152712, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "10e933efb739aa29898f0d6ec85216a1e3a887f656ea440e3dfb73055489427a", "compiled_ids_sha256": "b44559e03bdf5fd54bda7ce3b93240c4bfc7ef936bc36685d683b0f19bbde4d8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.75, 13.4375, 12.0, 12.1875, 12.25, 19.25], "runtime_probabilities": [0.0014930296456441283, 0.0029692440293729305, 0.0007052572909742594, 0.0008507026359438896, 0.0009055683040060103, 0.9930762052536011], "decide_seconds": 0.17265191600017715, "probabilities_f64_minus_runtime_max_abs": 7.219265518720874e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:04Z"} {"id": "c-darwin-01-main-02", "uid": "c010", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "prompt_sha256_by_order": ["b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285"], "request_sha256": "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "compiled_prompt_sha256_by_order": ["845aaf3e77ec7fd9ccefeb0369bad6914f74a0a4e47c5038bcac81310111f0a8", "c0f56b4a3afc646003c0ad4d6a0bf9b4c61e37dcea864a1b30612c7c4b598c46", "dad99b7d8ff8c4ed85f54efc8b59d31ce90010731bc3f3d998d8822eb7f3fff5", "711cb74e2e74eee811b11c2a3657aaad3418f80fda1edf8aafc2f409485d5561", "c6a195b28e61cccd7241e6669c72ef9c4393c217ebfc859cb0ebb54828c32650", "60b895f7526886c0f0e93a5b2c1e6b89b928d31850472819195ecd2731436637"], "compiled_tokens_by_order": [401, 401, 401, 401, 401, 401], "choice": "darwin", "probs": {"darwin": 0.5266753351739607, "einstein": 0.015904217035326202, "hypatia": 0.1175171518835938, "ibn_sina": 0.15089500991528815, "sagan": 0.018021838930818948, "socrates": 0.17098644706101218}, "confidence": 0.26613269563552455, "seconds": 0.1770316835002935, "seconds_all_orders": 1.062190101001761, "seconds_by_order": [0.17709933199876104, 0.17672584400133928, 0.17727655000271625, 0.1771371309987444, 0.17697040200073388, 0.17698084199946607], "argmax_stable": false, "spread": {"darwin": 0.525502802663003, "einstein": 0.04532819185981288, "hypatia": 0.08931840031890387, "ibn_sina": 0.3555818916766286, "sagan": 0.01937640629580853, "socrates": 0.1506716526198156}, "spread_2dp": {"darwin": 0.52, "einstein": 0.049999999999999996, "hypatia": 0.09, "ibn_sina": 0.35, "sagan": 0.019999999999999997, "socrates": 0.15000000000000002}, "choices_by_order": ["darwin", "darwin", "ibn_sina", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.5266753351739607, "einstein": 0.015904217035326202, "hypatia": 0.1175171518835938, "ibn_sina": 0.15089500991528815, "sagan": 0.018021838930818948, "socrates": 0.17098644706101218}, {"ibn_sina": 0.1524410107114583, "hypatia": 0.028198751564689924, "sagan": 0.008079077588752455, "socrates": 0.07200803472612766, "darwin": 0.6831932115768306, "einstein": 0.056079913832141136}, {"hypatia": 0.08553670669015993, "darwin": 0.3383042375855364, "sagan": 0.02031674863577474, "ibn_sina": 0.38334892348600264, "einstein": 0.03146719585682161, "socrates": 0.14102618774570466}, {"ibn_sina": 0.08456884405642223, "einstein": 0.010751721972328262, "darwin": 0.7080862619738441, "sagan": 0.027455483884560986, "hypatia": 0.08456884405642223, "socrates": 0.08456884405642223}, {"einstein": 0.038608661602279284, "sagan": 0.015119369121913044, "socrates": 0.026535319187715227, "ibn_sina": 0.07213047596032694, "darwin": 0.7754756981674384, "hypatia": 0.07213047596032694}, {"socrates": 0.02031479444119658, "ibn_sina": 0.027767031809374017, "hypatia": 0.06257397212008006, "einstein": 0.01396214042027685, "sagan": 0.01157502096053304, "darwin": 0.8638070402485394}], "prompt_tokens": 401, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.5266753351739607, "einstein": 0.015904217035326202, "hypatia": 0.1175171518835938, "ibn_sina": 0.15089500991528815, "sagan": 0.018021838930818948, "socrates": 0.17098644706101218}, "confidence": 0.26613269563552455, "choice": "darwin"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 177.06683199867257, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "845aaf3e77ec7fd9ccefeb0369bad6914f74a0a4e47c5038bcac81310111f0a8", "compiled_ids_sha256": "2c82207f9a93c3dc26bdba65b250a344432061dc02ca1018c708ad8067c3121f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 15.625, 17.625, 17.875, 15.75, 18.0], "runtime_probabilities": [0.5266753435134888, 0.015904216095805168, 0.1175171509385109, 0.15089501440525055, 0.01802183873951435, 0.17098644375801086], "decide_seconds": 0.172296988999733, "probabilities_f64_minus_runtime_max_abs": 8.339528068113111e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.1524410107114583, "hypatia": 0.028198751564689924, "sagan": 0.008079077588752455, "socrates": 0.07200803472612766, "darwin": 0.6831932115768306, "einstein": 0.056079913832141136}, "confidence": 0.4209092253802257, "choice": "darwin"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 176.69245400247746, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "c0f56b4a3afc646003c0ad4d6a0bf9b4c61e37dcea864a1b30612c7c4b598c46", "compiled_ids_sha256": "885b49844a750f95496e2fcd51c3a0a038a50cfd500fcde03ed2de46a8309fec", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.625, 15.9375, 14.6875, 16.875, 19.125, 16.625], "runtime_probabilities": [0.15244100987911224, 0.02819875255227089, 0.008079077117145061, 0.07200802862644196, 0.6831932067871094, 0.056079912930727005], "decide_seconds": 0.17206402999727288, "probabilities_f64_minus_runtime_max_abs": 6.099685700977631e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.08553670669015993, "darwin": 0.3383042375855364, "sagan": 0.02031674863577474, "ibn_sina": 0.38334892348600264, "einstein": 0.03146719585682161, "socrates": 0.14102618774570466}, "confidence": 0.2137463663009468, "choice": "ibn_sina"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 177.24919999818667, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "dad99b7d8ff8c4ed85f54efc8b59d31ce90010731bc3f3d998d8822eb7f3fff5", "compiled_ids_sha256": "656aeb0145a2ecafd9b7f8fad09ab24bea6eef2b61e89c8eb9f4ca7b83b8f6c2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 18.375, 15.5625, 18.5, 16.0, 17.5], "runtime_probabilities": [0.08553671091794968, 0.3383042514324188, 0.020316749811172485, 0.3833489418029785, 0.03146719932556152, 0.14102621376514435], "decide_seconds": 0.17266249599924777, "probabilities_f64_minus_runtime_max_abs": 2.6019439686164603e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.08456884405642223, "einstein": 0.010751721972328262, "darwin": 0.7080862619738441, "sagan": 0.027455483884560986, "hypatia": 0.08456884405642223, "socrates": 0.08456884405642223}, "confidence": 0.43152587693785605, "choice": "darwin"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 177.1098709978105, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "711cb74e2e74eee811b11c2a3657aaad3418f80fda1edf8aafc2f409485d5561", "compiled_ids_sha256": "48f156de0c08b8519effc342a4da6257c28bcace39978846d0ed7b62aa58fdd6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 15.1875, 19.375, 16.125, 17.25, 17.25], "runtime_probabilities": [0.08456884324550629, 0.010751721449196339, 0.7080862522125244, 0.027455482631921768, 0.08456884324550629, 0.08456884324550629], "decide_seconds": 0.17236270699868328, "probabilities_f64_minus_runtime_max_abs": 9.761319641476973e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.038608661602279284, "sagan": 0.015119369121913044, "socrates": 0.026535319187715227, "ibn_sina": 0.07213047596032694, "darwin": 0.7754756981674384, "hypatia": 0.07213047596032694}, "confidence": 0.5190126730524771, "choice": "darwin"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 176.94219300028635, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "c6a195b28e61cccd7241e6669c72ef9c4393c217ebfc859cb0ebb54828c32650", "compiled_ids_sha256": "b062188d673891c8e1dfc1c7ebc45971b69840fbb55bfe5feacc5b0bcedeb8f9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 15.4375, 16.0, 17.0, 19.375, 17.0], "runtime_probabilities": [0.03860866278409958, 0.015119371004402637, 0.026535319164395332, 0.0721304789185524, 0.7754757404327393, 0.0721304789185524], "decide_seconds": 0.1722675689998141, "probabilities_f64_minus_runtime_max_abs": 4.2265300814747775e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.02031479444119658, "ibn_sina": 0.027767031809374017, "hypatia": 0.06257397212008006, "einstein": 0.01396214042027685, "sagan": 0.01157502096053304, "darwin": 0.8638070402485394}, "confidence": 0.6708245089611506, "choice": "darwin"}}, "usage": {"input_tokens": 401, "output_tokens": 1}, "latency_ms": 176.95327199908206, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 401, "compiled_prompt_sha256": "60b895f7526886c0f0e93a5b2c1e6b89b928d31850472819195ecd2731436637", "compiled_ids_sha256": "5b63a165937f644ee727fd04f307c0fc69c779463d881b0bf2751da3d83761c6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 15.8125, 16.625, 15.125, 14.9375, 19.25], "runtime_probabilities": [0.020314795896410942, 0.027767036110162735, 0.06257397681474686, 0.01396214123815298, 0.011575022712349892, 0.8638071417808533], "decide_seconds": 0.17237998800192145, "probabilities_f64_minus_runtime_max_abs": 1.015323138897628e-07}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:05Z"} {"id": "c-darwin-01-main-03", "uid": "c011", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "prompt_sha256_by_order": ["c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5"], "request_sha256": "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "compiled_prompt_sha256_by_order": ["67ed85f77269ae70491a1864dafa1c40fd5c3c6a2590a97dc9bb1b6af2efd7d9", "d52ddc7906085a25361cdef95ad18623fd050948d5f3ee668e7c6773611100d9", "4dd4b32c9d73f073a47034e6e6e747f7eb03445fd051cbc33402493a06c5475a", "28b60c3fc9331143d1ccf39bc628a8f9092336156e1fb92ca3dadee0363bf83a", "2db636df227dd59cce1576cba21f762a6d76822ed2d9d94692fe9f89b0d7e44a", "d59853cbf40844109510a84c212219d3b41cec000aa6123862bd5ad5559439a5"], "compiled_tokens_by_order": [402, 402, 402, 402, 402, 402], "choice": "darwin", "probs": {"darwin": 0.9956514843661941, "einstein": 0.0008529087445829299, "hypatia": 0.0011657880324548594, "ibn_sina": 0.0006240013676021309, "sagan": 0.0008529087445829299, "socrates": 0.0008529087445829299}, "confidence": 0.9805218937823226, "seconds": 0.18393186050040336, "seconds_all_orders": 1.1035911630024202, "seconds_by_order": [0.18644773900086875, 0.18392453900014516, 0.1840701680011989, 0.18541900699710823, 0.18468776300142054, 0.17904194700167864], "argmax_stable": true, "spread": {"darwin": 0.008671370124327615, "einstein": 0.0014465815372640372, "hypatia": 0.00100903488955703, "ibn_sina": 0.002719939097455737, "sagan": 0.0014651289761197731, "socrates": 0.00270669735142238}, "spread_2dp": {"darwin": 0.010000000000000009, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9956514843661941, "einstein": 0.0008529087445829299, "hypatia": 0.0011657880324548594, "ibn_sina": 0.0006240013676021309, "sagan": 0.0008529087445829299, "socrates": 0.0008529087445829299}, {"ibn_sina": 0.0033439404650578678, "hypatia": 0.0020282024163113185, "sagan": 0.001789882350209331, "socrates": 0.00355960609600531, "darwin": 0.9869801142418665, "einstein": 0.0022982544305495704}, {"hypatia": 0.0021748229220118894, "darwin": 0.9942088010937759, "sagan": 0.0007060616276986557, "ibn_sina": 0.0009650718215454501, "einstein": 0.000851672893285533, "socrates": 0.001093569641682611}, {"ibn_sina": 0.0016909249424510687, "einstein": 0.0010255978208693933, "darwin": 0.9925483060347249, "sagan": 0.0021711906038184288, "hypatia": 0.0014018260139494537, "socrates": 0.001162154584186861}, {"einstein": 0.0019163309167169686, "sagan": 0.0010918924428461934, "socrates": 0.0014020176489037796, "ibn_sina": 0.0013170736936752479, "darwin": 0.9926839911678301, "hypatia": 0.0015886941300277068}, {"socrates": 0.0016909032740323173, "ibn_sina": 0.0012370912623190895, "hypatia": 0.0021711627810180128, "einstein": 0.0009634476438248997, "sagan": 0.0014018080501993638, "darwin": 0.9925355869886063}], "prompt_tokens": 402, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9956514843661941, "einstein": 0.0008529087445829299, "hypatia": 0.0011657880324548594, "ibn_sina": 0.0006240013676021309, "sagan": 0.0008529087445829299, "socrates": 0.0008529087445829299}, "confidence": 0.9805218937823226, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 186.41783999919426, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "67ed85f77269ae70491a1864dafa1c40fd5c3c6a2590a97dc9bb1b6af2efd7d9", "compiled_ids_sha256": "f12e7fc6a162cdd7569eba4da42ac19c958d6ecfbd2351179f1c50f8bd12014a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 12.6875, 13.0, 12.375, 12.6875, 12.6875], "runtime_probabilities": [0.9956515431404114, 0.000852908764500171, 0.0011657880386337638, 0.0006240013753995299, 0.000852908764500171, 0.000852908764500171], "decide_seconds": 0.1813550690021657, "probabilities_f64_minus_runtime_max_abs": 5.8774217270851636e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0033439404650578678, "hypatia": 0.0020282024163113185, "sagan": 0.001789882350209331, "socrates": 0.00355960609600531, "darwin": 0.9869801142418665, "einstein": 0.0022982544305495704}, "confidence": 0.9498101520148856, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 183.89881899929605, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "d52ddc7906085a25361cdef95ad18623fd050948d5f3ee668e7c6773611100d9", "compiled_ids_sha256": "9278fdf813c8a5b0b95054149ed642ad320e34842969e3be93fb44b675f2832e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.9375, 13.4375, 13.3125, 14.0, 19.625, 13.5625], "runtime_probabilities": [0.003343940246850252, 0.002028202172368765, 0.0017898821970447898, 0.0035596059169620275, 0.9869800806045532, 0.0022982542868703604], "decide_seconds": 0.1789909869985422, "probabilities_f64_minus_runtime_max_abs": 3.3637313268108926e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0021748229220118894, "darwin": 0.9942088010937759, "sagan": 0.0007060616276986557, "ibn_sina": 0.0009650718215454501, "einstein": 0.000851672893285533, "socrates": 0.001093569641682611}, "confidence": 0.9752155132212778, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 184.04404800094198, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "4dd4b32c9d73f073a47034e6e6e747f7eb03445fd051cbc33402493a06c5475a", "compiled_ids_sha256": "2ba8d8484869807159429752d76b9332b0d3916ce4545ab7a943d6a6eb7d05d1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 19.625, 12.375, 12.6875, 12.5625, 12.8125], "runtime_probabilities": [0.0021748230792582035, 0.994208812713623, 0.0007060616044327617, 0.0009650717838667333, 0.0008516728994436562, 0.0010935695609077811], "decide_seconds": 0.1791086869998253, "probabilities_f64_minus_runtime_max_abs": 1.16198471955542e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0016909249424510687, "einstein": 0.0010255978208693933, "darwin": 0.9925483060347249, "sagan": 0.0021711906038184288, "hypatia": 0.0014018260139494537, "socrates": 0.001162154584186861}, "confidence": 0.9689395825008651, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 185.39274700015085, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "28b60c3fc9331143d1ccf39bc628a8f9092336156e1fb92ca3dadee0363bf83a", "compiled_ids_sha256": "f47ed8b1971e6c2f470e0da907f26d2fe1530e352f7329a2b51acf4440bd1ba4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 12.875, 19.75, 13.625, 13.1875, 13.0], "runtime_probabilities": [0.0016909249825403094, 0.0010255977977067232, 0.9925482869148254, 0.002171190455555916, 0.0014018260408192873, 0.0011621544836089015], "decide_seconds": 0.18039806600063457, "probabilities_f64_minus_runtime_max_abs": 1.9119899441477628e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0019163309167169686, "sagan": 0.0010918924428461934, "socrates": 0.0014020176489037796, "ibn_sina": 0.0013170736936752479, "darwin": 0.9926839911678301, "hypatia": 0.0015886941300277068}, "confidence": 0.9693529957477297, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 184.65947299773688, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "2db636df227dd59cce1576cba21f762a6d76822ed2d9d94692fe9f89b0d7e44a", "compiled_ids_sha256": "431e0ab2d134a50837ac255a8b84534c15d958ed6646d807dc372a5d2b6a1917", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 12.8125, 13.0625, 13.0, 19.625, 13.1875], "runtime_probabilities": [0.001916330773383379, 0.0010918923653662205, 0.0014020175440236926, 0.0013170735910534859, 0.9926839470863342, 0.0015886940527707338], "decide_seconds": 0.17972749099862995, "probabilities_f64_minus_runtime_max_abs": 4.408149589796295e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0016909032740323173, "ibn_sina": 0.0012370912623190895, "hypatia": 0.0021711627810180128, "einstein": 0.0009634476438248997, "sagan": 0.0014018080501993638, "darwin": 0.9925355869886063}, "confidence": 0.9688983796688898, "choice": "darwin"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 179.01673700180254, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "d59853cbf40844109510a84c212219d3b41cec000aa6123862bd5ad5559439a5", "compiled_ids_sha256": "b810e433af9e97e7b77d389f176bd2d0fe6591ea3d97424584348614ea4ef64c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 12.9375, 13.5, 12.6875, 13.0625, 19.625], "runtime_probabilities": [0.0016909032128751278, 0.0012370911426842213, 0.0021711625158786774, 0.000963447499088943, 0.0014018078800290823, 0.9925354719161987], "decide_seconds": 0.17418566400010604, "probabilities_f64_minus_runtime_max_abs": 1.1507240760266768e-07}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:06Z"} {"id": "c-darwin-01-starter-01", "uid": "c012", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "prompt_sha256_by_order": ["97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1"], "request_sha256": "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "compiled_prompt_sha256_by_order": ["9e3f59abe3f3521d1414d75abb72215d06b06fabcaa7452231957e31bc3997a0", "8d8d52d24ae0d69da1b42ca2bf2d9f23899f777722a968bf55dc653738e402ad", "818369bbaf473d78c1e70dc8e86a92ae4c14cf58b91d017b51948f1c911e0516", "0c3bf7de2e1d0269fd25c830086d150e4bdfa13a669c8e7d6419b0f53713ac09", "244a5833bbffa916ed9d6646487ae937bbd4fe94a9654b6e61b8c6ae32278738", "ef53f77a12c03833d222d2c68857c7aa1bf109028bd05604056b23ae46da6e67"], "compiled_tokens_by_order": [240, 240, 240, 240, 240, 240], "choice": "sagan", "probs": {"darwin": 0.03383205421089814, "einstein": 0.01244611719678715, "hypatia": 0.046242935323304285, "ibn_sina": 0.020520178159969628, "sagan": 0.8196756412268883, "socrates": 0.06728307388215249}, "confidence": 0.5894378356163983, "seconds": 0.15902832866595418, "seconds_all_orders": 0.9541699719957251, "seconds_by_order": [0.16297941000084393, 0.15701262499715085, 0.15728324399970006, 0.16290724099962972, 0.1571511649999593, 0.1568362869984412], "argmax_stable": false, "spread": {"darwin": 0.25463965861318716, "einstein": 0.18910249818219405, "hypatia": 0.12934085880982246, "ibn_sina": 0.2944718936487917, "sagan": 0.7408254711123634, "socrates": 0.23505666359052657}, "spread_2dp": {"darwin": 0.26, "einstein": 0.19, "hypatia": 0.12, "ibn_sina": 0.29, "sagan": 0.74, "socrates": 0.22999999999999998}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "ibn_sina", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.03383205421089814, "einstein": 0.01244611719678715, "hypatia": 0.046242935323304285, "ibn_sina": 0.020520178159969628, "sagan": 0.8196756412268883, "socrates": 0.06728307388215249}, {"ibn_sina": 0.037282745022182776, "hypatia": 0.025624030937643708, "sagan": 0.5478659387357658, "socrates": 0.12224541464998126, "darwin": 0.06543325527544527, "einstein": 0.2015486153789812}, {"hypatia": 0.04091723400454815, "darwin": 0.020574477340713188, "sagan": 0.30233973747267906, "ibn_sina": 0.2077948601174495, "einstein": 0.126033953591931, "socrates": 0.30233973747267906}, {"ibn_sina": 0.04101667181198543, "einstein": 0.030008438145023714, "darwin": 0.08683229490736978, "sagan": 0.5662176888941199, "hypatia": 0.06762505926974675, "socrates": 0.20829984697175444}, {"einstein": 0.11472632055781849, "sagan": 0.07885017011452494, "socrates": 0.13000195266614084, "ibn_sina": 0.31185847241828535, "darwin": 0.2752141359539003, "hypatia": 0.08934894828933004}, {"socrates": 0.15496488974746617, "ibn_sina": 0.017386578769493597, "hypatia": 0.15496488974746617, "einstein": 0.025297322839196276, "sagan": 0.5408806115839192, "darwin": 0.10650570731245855}], "prompt_tokens": 240, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.03383205421089814, "einstein": 0.01244611719678715, "hypatia": 0.046242935323304285, "ibn_sina": 0.020520178159969628, "sagan": 0.8196756412268883, "socrates": 0.06728307388215249}, "confidence": 0.5894378356163983, "choice": "sagan"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 162.95476999948733, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "9e3f59abe3f3521d1414d75abb72215d06b06fabcaa7452231957e31bc3997a0", "compiled_ids_sha256": "79ebf5af4c3e548555cf057c1040dde2bbc5c096909a258efb54835918c41f5f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 14.8125, 16.125, 15.3125, 19.0, 16.5], "runtime_probabilities": [0.03383205458521843, 0.012446116656064987, 0.046242937445640564, 0.020520178601145744, 0.819675624370575, 0.06728307157754898], "decide_seconds": 0.15984077399843954, "probabilities_f64_minus_runtime_max_abs": 1.6856313345137153e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.037282745022182776, "hypatia": 0.025624030937643708, "sagan": 0.5478659387357658, "socrates": 0.12224541464998126, "darwin": 0.06543325527544527, "einstein": 0.2015486153789812}, "confidence": 0.2720239192171856, "choice": "sagan"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 156.99186499841744, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "8d8d52d24ae0d69da1b42ca2bf2d9f23899f777722a968bf55dc653738e402ad", "compiled_ids_sha256": "0a1d8f77807eb4997b7dfd9a04b1578f84abfe58e2bdcfdff3d88729806a2f3b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.4375, 18.5, 17.0, 16.375, 17.5], "runtime_probabilities": [0.03728274255990982, 0.025624031201004982, 0.5478659272193909, 0.1222454160451889, 0.06543325632810593, 0.20154863595962524], "decide_seconds": 0.15403563899963046, "probabilities_f64_minus_runtime_max_abs": 2.0580644050083308e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.04091723400454815, "darwin": 0.020574477340713188, "sagan": 0.30233973747267906, "ibn_sina": 0.2077948601174495, "einstein": 0.126033953591931, "socrates": 0.30233973747267906}, "confidence": 0.1508150958259422, "choice": "sagan"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 157.26239400100894, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "818369bbaf473d78c1e70dc8e86a92ae4c14cf58b91d017b51948f1c911e0516", "compiled_ids_sha256": "303d303aa6647d263a2d077c744c809437adca7e2487101ea586dcf55344d142", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 15.4375, 18.125, 17.75, 17.25, 18.125], "runtime_probabilities": [0.04091723635792732, 0.020574478432536125, 0.3023397624492645, 0.20779487490653992, 0.1260339617729187, 0.3023397624492645], "decide_seconds": 0.15434667600129615, "probabilities_f64_minus_runtime_max_abs": 2.4976585466518486e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04101667181198543, "einstein": 0.030008438145023714, "darwin": 0.08683229490736978, "sagan": 0.5662176888941199, "hypatia": 0.06762505926974675, "socrates": 0.20829984697175444}, "confidence": 0.2859483157365055, "choice": "sagan"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 162.88546100258827, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "0c3bf7de2e1d0269fd25c830086d150e4bdfa13a669c8e7d6419b0f53713ac09", "compiled_ids_sha256": "45f0ff29b7458d0cd0e0cffa3265d3dfe1509bf4f5af99b940c9a55b9010b315", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 15.4375, 16.5, 18.375, 16.25, 17.375], "runtime_probabilities": [0.04101667180657387, 0.030008437111973763, 0.08683229237794876, 0.5662176609039307, 0.06762505322694778, 0.20829986035823822], "decide_seconds": 0.15970369500064407, "probabilities_f64_minus_runtime_max_abs": 2.799018927746033e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.11472632055781849, "sagan": 0.07885017011452494, "socrates": 0.13000195266614084, "ibn_sina": 0.31185847241828535, "darwin": 0.2752141359539003, "hypatia": 0.08934894828933004}, "confidence": 0.08012721470659201, "choice": "ibn_sina"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 157.12953500042204, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "244a5833bbffa916ed9d6646487ae937bbd4fe94a9654b6e61b8c6ae32278738", "compiled_ids_sha256": "9cac17e1afd0057ff1a4a8f232264d2dd594a839a4280ed64b036ad597fea0fd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 16.625, 17.125, 18.0, 17.875, 16.75], "runtime_probabilities": [0.11472633481025696, 0.0788501724600792, 0.1300019472837448, 0.31185847520828247, 0.27521413564682007, 0.08934894949197769], "decide_seconds": 0.1540658779995283, "probabilities_f64_minus_runtime_max_abs": 1.4252438465534034e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.15496488974746617, "ibn_sina": 0.017386578769493597, "hypatia": 0.15496488974746617, "einstein": 0.025297322839196276, "sagan": 0.5408806115839192, "darwin": 0.10650570731245855}, "confidence": 0.26760252004575735, "choice": "sagan"}}, "usage": {"input_tokens": 240, "output_tokens": 1}, "latency_ms": 156.8159470007231, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 240, "compiled_prompt_sha256": "ef53f77a12c03833d222d2c68857c7aa1bf109028bd05604056b23ae46da6e67", "compiled_ids_sha256": "7bbba6d95257e9825d1aeefda8a77a4e758aec3f53b58511dc0facd4f924e1e9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.75, 15.5625, 17.75, 15.9375, 19.0, 17.375], "runtime_probabilities": [0.15496490895748138, 0.017386581748723984, 0.15496490895748138, 0.02529732696712017, 0.5408806800842285, 0.10650571435689926], "decide_seconds": 0.15388287899986608, "probabilities_f64_minus_runtime_max_abs": 6.850030931371265e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:07Z"} {"id": "c-darwin-01-starter-01", "uid": "c013", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "prompt_sha256_by_order": ["b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511"], "request_sha256": "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "compiled_prompt_sha256_by_order": ["4503591e6f0e5386c6e8577056733fa060b0d80509ece075b927a593b418dd73", "c99d99efe7d0d9d15d27bb1cf2e7d32fec3f3339cbfc311efe0bf04fb1140bf1", "486326b0f3d7561f7480e438d8bc8483980de1a4517534b089339a49924bfc19", "be9a9ea0c51aa522dde769357e4eeab2be1586c5acd8003b882c6f54d5df37a0", "23be69ba5cb1f451624fd3b7c40ae82e0b4db49b8ca665f6aeb2c3876ffe12a3", "07e98ff80964b0a6def8b7573c24437abd99e5a530b4fe072fb31b28b2c86dd1"], "compiled_tokens_by_order": [269, 269, 269, 269, 269, 269], "choice": "darwin", "probs": {"darwin": 0.9936200948773601, "einstein": 0.0014938473935687173, "hypatia": 0.0015901922729227864, "ibn_sina": 0.0010929221003960787, "sagan": 0.0012384429873863632, "socrates": 0.0009645003683657897}, "confidence": 0.9727831464324574, "seconds": 0.16885966283310458, "seconds_all_orders": 1.0131579769986274, "seconds_by_order": [0.1679390820027038, 0.17193453099753242, 0.1652100219980639, 0.17273470500003896, 0.17090134899990517, 0.16443828800038318], "argmax_stable": true, "spread": {"darwin": 0.003781419419908949, "einstein": 0.0018672791302815621, "hypatia": 0.0012072739281590483, "ibn_sina": 0.001365417825315763, "sagan": 0.0011462685261035429, "socrates": 0.0009535502511176387}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9936200948773601, "einstein": 0.0014938473935687173, "hypatia": 0.0015901922729227864, "ibn_sina": 0.0010929221003960787, "sagan": 0.0012384429873863632, "socrates": 0.0009645003683657897}, {"ibn_sina": 0.0024583399257118417, "hypatia": 0.001236130944034651, "sagan": 0.001161237556174105, "socrates": 0.0016895906745655438, "darwin": 0.9917651102249484, "einstein": 0.0016895906745655438}, {"hypatia": 0.002171211983110826, "darwin": 0.9925580794652176, "sagan": 0.0013169066364921843, "ibn_sina": 0.001401839817458086, "einstein": 0.0015884926206001123, "socrates": 0.000963469477121236}, {"ibn_sina": 0.0023075060822776478, "einstein": 0.0020363669703051383, "darwin": 0.9909532149386943, "sagan": 0.0023075060822776478, "hypatia": 0.0012351190019161923, "socrates": 0.0011602869245292136}, {"einstein": 0.0029595640461024145, "sagan": 0.0016863087621603132, "socrates": 0.0016863087621603132, "ibn_sina": 0.001795066333344288, "darwin": 0.9898386754574512, "hypatia": 0.00203407663878142}, {"socrates": 0.0019170197282388746, "ibn_sina": 0.0014929764655157672, "hypatia": 0.0009639380549517777, "einstein": 0.0010922849158208523, "sagan": 0.0014929764655157672, "darwin": 0.9930408043699569}], "prompt_tokens": 269, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9936200948773601, "einstein": 0.0014938473935687173, "hypatia": 0.0015901922729227864, "ibn_sina": 0.0010929221003960787, "sagan": 0.0012384429873863632, "socrates": 0.0009645003683657897}, "confidence": 0.9727831464324574, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 167.91219200240448, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "4503591e6f0e5386c6e8577056733fa060b0d80509ece075b927a593b418dd73", "compiled_ids_sha256": "fc3048d46a00e63f863e4d4d211b3e8a6c309b8b184a6adb0f036ee0575056f9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 12.875, 12.9375, 12.5625, 12.6875, 12.4375], "runtime_probabilities": [0.9936200976371765, 0.0014938473468646407, 0.0015901923179626465, 0.0010929220588877797, 0.0012384430738165975, 0.0009645003592595458], "decide_seconds": 0.16448958799810498, "probabilities_f64_minus_runtime_max_abs": 2.759816375430546e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0024583399257118417, "hypatia": 0.001236130944034651, "sagan": 0.001161237556174105, "socrates": 0.0016895906745655438, "darwin": 0.9917651102249484, "einstein": 0.0016895906745655438}, "confidence": 0.9661415024328073, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 171.91024100247887, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "c99d99efe7d0d9d15d27bb1cf2e7d32fec3f3339cbfc311efe0bf04fb1140bf1", "compiled_ids_sha256": "1922fb71f79b3713cc991563e5534f816800e5294bf27df14f7752e21b49372d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 12.5625, 12.5, 12.875, 19.25, 12.875], "runtime_probabilities": [0.002458340022712946, 0.0012361309491097927, 0.0011612374801188707, 0.001689590746536851, 0.9917650818824768, 0.001689590746536851], "decide_seconds": 0.1685205270005099, "probabilities_f64_minus_runtime_max_abs": 2.8342471591891183e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.002171211983110826, "darwin": 0.9925580794652176, "sagan": 0.0013169066364921843, "ibn_sina": 0.001401839817458086, "einstein": 0.0015884926206001123, "socrates": 0.000963469477121236}, "confidence": 0.9689676571308761, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 165.18628300036653, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "486326b0f3d7561f7480e438d8bc8483980de1a4517534b089339a49924bfc19", "compiled_ids_sha256": "b68d59f7901b3a45c5da3d4a2126460edcf292f6c42d7730720029352f40c6c2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 19.625, 13.0, 13.0625, 13.1875, 12.6875], "runtime_probabilities": [0.002171211875975132, 0.9925580024719238, 0.0013169065350666642, 0.0014018397778272629, 0.0015884925378486514, 0.0009634693851694465], "decide_seconds": 0.16176860899940948, "probabilities_f64_minus_runtime_max_abs": 7.699329374055708e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0023075060822776478, "einstein": 0.0020363669703051383, "darwin": 0.9909532149386943, "sagan": 0.0023075060822776478, "hypatia": 0.0012351190019161923, "socrates": 0.0011602869245292136}, "confidence": 0.9632996036638588, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 172.711335002532, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "be9a9ea0c51aa522dde769357e4eeab2be1586c5acd8003b882c6f54d5df37a0", "compiled_ids_sha256": "ca43b1caf0e65a626ec0916d80015aa8d074aef928db46725f365f5fce7597cd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 13.1875, 19.375, 13.3125, 12.6875, 12.625], "runtime_probabilities": [0.0023075060453265905, 0.002036366844549775, 0.9909532070159912, 0.0023075060453265905, 0.001235119067132473, 0.001160286832600832], "decide_seconds": 0.16923841199968592, "probabilities_f64_minus_runtime_max_abs": 7.922703049167978e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0029595640461024145, "sagan": 0.0016863087621603132, "socrates": 0.0016863087621603132, "ibn_sina": 0.001795066333344288, "darwin": 0.9898386754574512, "hypatia": 0.00203407663878142}, "confidence": 0.9593508991394184, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 170.87745900062146, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "23be69ba5cb1f451624fd3b7c40ae82e0b4db49b8ca665f6aeb2c3876ffe12a3", "compiled_ids_sha256": "3b3b7a4bfea2702a58446afebffef4fbca608d940bd17511298ea5204d64d231", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.4375, 12.875, 12.875, 12.9375, 19.25, 13.0625], "runtime_probabilities": [0.0029595643281936646, 0.0016863089986145496, 0.0016863089986145496, 0.0017950664041563869, 0.989838719367981, 0.0020340767223387957], "decide_seconds": 0.16734037599962903, "probabilities_f64_minus_runtime_max_abs": 4.391052976782106e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0019170197282388746, "ibn_sina": 0.0014929764655157672, "hypatia": 0.0009639380549517777, "einstein": 0.0010922849158208523, "sagan": 0.0014929764655157672, "darwin": 0.9930408043699569}, "confidence": 0.9706980187534671, "choice": "darwin"}}, "usage": {"input_tokens": 269, "output_tokens": 1}, "latency_ms": 164.41560800012667, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 269, "compiled_prompt_sha256": "07e98ff80964b0a6def8b7573c24437abd99e5a530b4fe072fb31b28b2c86dd1", "compiled_ids_sha256": "7f03f77ae1e8fc1b0036d763bbb74bb88191536eb1211803850576b5e3fc0582", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.125, 12.875, 12.4375, 12.5625, 12.875, 19.375], "runtime_probabilities": [0.0019170197192579508, 0.0014929764438420534, 0.0009639380150474608, 0.001092284917831421, 0.0014929764438420534, 0.9930408000946045], "decide_seconds": 0.16106780499831075, "probabilities_f64_minus_runtime_max_abs": 4.275352405436195e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:08Z"} {"id": "c-darwin-01-starter-01", "uid": "c014", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "prompt_sha256_by_order": ["9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564"], "request_sha256": "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "compiled_prompt_sha256_by_order": ["5d90f70b3e510382c9f8fe500bb3610f93699a121cbbf98eabe2725ae32a0559", "c50296a587f94f3042fa25945e6694ba84f7209fa29ab299e32c96f7a7792ba9", "21abcebe34b503bfd4819717c520c0bbfacba4a5ed4c68caa6c68a8c6c46a37e", "152457b9222c9769845f06b7dec35e359f8481e3af967922eb5385ecdff51e5f", "731204adf744e802b7b46e492bcbff22e53b2d27a7384a2a31ba5780b3312552", "3697ea310d117c069d18786e9e9dd55893aa44fbfba134fe57fd4ff3a64ebacb"], "compiled_tokens_by_order": [294, 294, 294, 294, 294, 294], "choice": "sagan", "probs": {"darwin": 0.07852623814818317, "einstein": 0.02113505027702185, "hypatia": 0.03273460623220716, "ibn_sina": 0.004715867153057001, "sagan": 0.8442366217840893, "socrates": 0.018651616405441475}, "confidence": 0.6451989738644032, "seconds": 0.1681137569994462, "seconds_all_orders": 1.0086825419966772, "seconds_by_order": [0.1721517389996734, 0.17104372800167766, 0.16516285299803712, 0.1651998730012565, 0.16601544599689078, 0.16910890299914172], "argmax_stable": true, "spread": {"darwin": 0.1309816232929245, "einstein": 0.0268794353622084, "hypatia": 0.040968799985464086, "ibn_sina": 0.005691740816888948, "sagan": 0.15695120509630778, "socrates": 0.09038148299926661}, "spread_2dp": {"darwin": 0.13, "einstein": 0.03, "hypatia": 0.04, "ibn_sina": 0.01, "sagan": 0.15000000000000002, "socrates": 0.09}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.07852623814818317, "einstein": 0.02113505027702185, "hypatia": 0.03273460623220716, "ibn_sina": 0.004715867153057001, "sagan": 0.8442366217840893, "socrates": 0.018651616405441475}, {"ibn_sina": 0.009547639822702096, "hypatia": 0.006164419582391868, "sagan": 0.807379634839963, "socrates": 0.051614053237171394, "darwin": 0.08509718743917329, "einstein": 0.040197065078598264}, {"hypatia": 0.008077550737690516, "darwin": 0.021957059388720085, "sagan": 0.9336379227480152, "ibn_sina": 0.005909662363303858, "einstein": 0.013317629716389865, "socrates": 0.017100175045880545}, {"ibn_sina": 0.01040760796994595, "einstein": 0.0194439715163909, "darwin": 0.1529386826816446, "sagan": 0.7766867176517074, "hypatia": 0.01040760796994595, "socrates": 0.030115412210365292}, {"einstein": 0.03489418548012344, "sagan": 0.7941880009016732, "socrates": 0.10748165804514716, "ibn_sina": 0.005351196657385115, "darwin": 0.050770740293674704, "hypatia": 0.007314218621996372}, {"socrates": 0.0252285944365898, "ibn_sina": 0.006378787620069241, "hypatia": 0.047133219567855954, "einstein": 0.032394156483888256, "sagan": 0.8354563070502217, "darwin": 0.05340893484137504}], "prompt_tokens": 294, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.07852623814818317, "einstein": 0.02113505027702185, "hypatia": 0.03273460623220716, "ibn_sina": 0.004715867153057001, "sagan": 0.8442366217840893, "socrates": 0.018651616405441475}, "confidence": 0.6451989738644032, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 172.1272399990994, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "5d90f70b3e510382c9f8fe500bb3610f93699a121cbbf98eabe2725ae32a0559", "compiled_ids_sha256": "1360cb8f139ab7df74cce524630cc666c0ab3b9f2b63c789ee0517529cc2e1d3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 15.8125, 16.25, 14.3125, 19.5, 15.6875], "runtime_probabilities": [0.07852622866630554, 0.02113504894077778, 0.03273460268974304, 0.0047158668749034405, 0.8442365527153015, 0.01865161396563053], "decide_seconds": 0.16848544700042112, "probabilities_f64_minus_runtime_max_abs": 6.906878780021941e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.009547639822702096, "hypatia": 0.006164419582391868, "sagan": 0.807379634839963, "socrates": 0.051614053237171394, "darwin": 0.08509718743917329, "einstein": 0.040197065078598264}, "confidence": 0.5867866098734904, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 171.02171799706412, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "c50296a587f94f3042fa25945e6694ba84f7209fa29ab299e32c96f7a7792ba9", "compiled_ids_sha256": "ab5288f2086e36d2ce132d8350f718f23d1b8eda8cf283b107932631cf036edc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 14.875, 19.75, 17.0, 17.5, 16.75], "runtime_probabilities": [0.009547638706862926, 0.006164418999105692, 0.8073796033859253, 0.051614049822092056, 0.08509718626737595, 0.040197063237428665], "decide_seconds": 0.16742869500012603, "probabilities_f64_minus_runtime_max_abs": 3.145403770776767e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.008077550737690516, "darwin": 0.021957059388720085, "sagan": 0.9336379227480152, "ibn_sina": 0.005909662363303858, "einstein": 0.013317629716389865, "socrates": 0.017100175045880545}, "confidence": 0.8078468590557069, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 165.13955400296254, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "21abcebe34b503bfd4819717c520c0bbfacba4a5ed4c68caa6c68a8c6c46a37e", "compiled_ids_sha256": "ab09642479d25cc586ce7d8b4685f7d699f98eaf88b15cdd2c1229750d764a16", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 16.25, 20.0, 14.9375, 15.75, 16.0], "runtime_probabilities": [0.008077550679445267, 0.02195706032216549, 0.9336379170417786, 0.00590966222807765, 0.01331762969493866, 0.017100175842642784], "decide_seconds": 0.16153203100111568, "probabilities_f64_minus_runtime_max_abs": 5.706236594349434e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01040760796994595, "einstein": 0.0194439715163909, "darwin": 0.1529386826816446, "sagan": 0.7766867176517074, "hypatia": 0.01040760796994595, "socrates": 0.030115412210365292}, "confidence": 0.5755100332851443, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 165.17744299926562, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "152457b9222c9769845f06b7dec35e359f8481e3af967922eb5385ecdff51e5f", "compiled_ids_sha256": "97b3d68f937b9608483f5e205e4a89e2c99c250e36d783d5e81e3bb2e66acffc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.9375, 18.0, 19.625, 15.3125, 16.375], "runtime_probabilities": [0.01040760800242424, 0.019443972036242485, 0.15293867886066437, 0.7766867280006409, 0.01040760800242424, 0.030115412548184395], "decide_seconds": 0.16150045100221178, "probabilities_f64_minus_runtime_max_abs": 1.0348933487946965e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03489418548012344, "sagan": 0.7941880009016732, "socrates": 0.10748165804514716, "ibn_sina": 0.005351196657385115, "darwin": 0.050770740293674704, "hypatia": 0.007314218621996372}, "confidence": 0.5785686380470851, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 165.99214599773404, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "731204adf744e802b7b46e492bcbff22e53b2d27a7384a2a31ba5780b3312552", "compiled_ids_sha256": "d2281dcb59f7cde98cfaf6d15ff30e234f166bfc18a9ec2df55ab2c3157cfbe5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 19.625, 17.625, 14.625, 16.875, 14.9375], "runtime_probabilities": [0.0348941870033741, 0.7941880226135254, 0.10748165845870972, 0.005351196974515915, 0.05077074095606804, 0.007314218673855066], "decide_seconds": 0.16238425399933476, "probabilities_f64_minus_runtime_max_abs": 2.1711852205008597e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0252285944365898, "ibn_sina": 0.006378787620069241, "hypatia": 0.047133219567855954, "einstein": 0.032394156483888256, "sagan": 0.8354563070502217, "darwin": 0.05340893484137504}, "confidence": 0.6166687016303505, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 169.08674300066195, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "3697ea310d117c069d18786e9e9dd55893aa44fbfba134fe57fd4ff3a64ebacb", "compiled_ids_sha256": "2f65e8c958cef8239d6109a8964eda844c5b24bd8ffb8d3f789ab6f2bfa614d4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 14.75, 16.75, 16.375, 19.625, 16.875], "runtime_probabilities": [0.02522859536111355, 0.006378787104040384, 0.047133222222328186, 0.032394155859947205, 0.8354563117027283, 0.05340893194079399], "decide_seconds": 0.16540794100001222, "probabilities_f64_minus_runtime_max_abs": 4.652506602198514e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:09Z"} {"id": "c-darwin-01-starter-02", "uid": "c015", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "prompt_sha256_by_order": ["30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb"], "request_sha256": "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "compiled_prompt_sha256_by_order": ["cc1df5806cd5c60ea952970a2eda3feb0945e82bb626caee1c5af7534e8df1ee", "52148491ffe991ea9c95419de43cd803e23eb70d293c15b71d16ccbdda347391", "fab59394c58f9dd65f213d2fe89491615f1c92b5ef3dbf6bbfec59942cb9f7ba", "3aeab417b11636c4046dba4e7af465fe2b636584ded218286c084cf35db3f09d", "f9ff1c68587c98be1aafb24549421ccc60d759a186e65e7d9c32de9712e17b3a", "ec8e25a98b9fbce7af2926d67b2dc722125fcc540a8b24ac2e530d7389875c2e"], "compiled_tokens_by_order": [273, 273, 273, 273, 273, 273], "choice": "darwin", "probs": {"darwin": 0.9204511627896921, "einstein": 0.01685865111237148, "hypatia": 0.009605769851960138, "ibn_sina": 0.019103354428777072, "sagan": 0.019103354428777072, "socrates": 0.014877707388422173}, "confidence": 0.7747611713866789, "seconds": 0.1661009141668425, "seconds_all_orders": 0.9966054850010551, "seconds_by_order": [0.16920074200243107, 0.16442303900112165, 0.16426296000281582, 0.16495290499733528, 0.16904253299799166, 0.16472330599935958], "argmax_stable": true, "spread": {"darwin": 0.1200455690988429, "einstein": 0.09025635236803238, "hypatia": 0.00844065807169404, "ibn_sina": 0.061308665396073825, "sagan": 0.034441146146096786, "socrates": 0.09514762525752245}, "spread_2dp": {"darwin": 0.12, "einstein": 0.09000000000000001, "hypatia": 0.01, "ibn_sina": 0.060000000000000005, "sagan": 0.03, "socrates": 0.1}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9204511627896921, "einstein": 0.01685865111237148, "hypatia": 0.009605769851960138, "ibn_sina": 0.019103354428777072, "sagan": 0.019103354428777072, "socrates": 0.014877707388422173}, {"ibn_sina": 0.018591711397103725, "hypatia": 0.006035844977484077, "sagan": 0.025411856528914666, "socrates": 0.060959874801049876, "darwin": 0.8415251140644552, "einstein": 0.04747559823099245}, {"hypatia": 0.010725441090678973, "darwin": 0.8004055936908492, "sagan": 0.03103509283944183, "ibn_sina": 0.0744493174585122, "einstein": 0.017683262863843606, "socrates": 0.06570129205667426}, {"ibn_sina": 0.013140652062438373, "einstein": 0.010233950116295876, "darwin": 0.8129833552243907, "sagan": 0.04047605788849216, "hypatia": 0.013140652062438373, "socrates": 0.11002533264594462}, {"einstein": 0.10049030248432826, "sagan": 0.006034911742395372, "socrates": 0.018588836830659497, "ibn_sina": 0.02879095369969152, "darwin": 0.8413950012521811, "hypatia": 0.0046999939907443325}, {"socrates": 0.026884255997327165, "ibn_sina": 0.02228783104240028, "hypatia": 0.008728040042625774, "einstein": 0.012699223328464383, "sagan": 0.039116361664508915, "darwin": 0.8902842879246734}], "prompt_tokens": 273, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9204511627896921, "einstein": 0.01685865111237148, "hypatia": 0.009605769851960138, "ibn_sina": 0.019103354428777072, "sagan": 0.019103354428777072, "socrates": 0.014877707388422173}, "confidence": 0.7747611713866789, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 169.17652299889596, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "cc1df5806cd5c60ea952970a2eda3feb0945e82bb626caee1c5af7534e8df1ee", "compiled_ids_sha256": "f791a198d080ca74c8294b34a536763716401dc9291fa42c4a1acd27960ce18f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 15.5, 14.9375, 15.625, 15.625, 15.375], "runtime_probabilities": [0.9204511642456055, 0.016858652234077454, 0.009605769999325275, 0.01910335384309292, 0.01910335384309292, 0.01487770676612854], "decide_seconds": 0.1657178589994146, "probabilities_f64_minus_runtime_max_abs": 1.4559133898472965e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.018591711397103725, "hypatia": 0.006035844977484077, "sagan": 0.025411856528914666, "socrates": 0.060959874801049876, "darwin": 0.8415251140644552, "einstein": 0.04747559823099245}, "confidence": 0.6323860255227054, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.40154899828485, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "52148491ffe991ea9c95419de43cd803e23eb70d293c15b71d16ccbdda347391", "compiled_ids_sha256": "aed46d4b36b34ae538fa0df4c24ada9c8eedf149a9a24196484367a9c22f3e5e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.4375, 15.875, 16.75, 19.375, 16.5], "runtime_probabilities": [0.018591711297631264, 0.006035844795405865, 0.025411857292056084, 0.06095987930893898, 0.841525137424469, 0.047475602477788925], "decide_seconds": 0.16105268399769557, "probabilities_f64_minus_runtime_max_abs": 2.336001381220143e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.010725441090678973, "darwin": 0.8004055936908492, "sagan": 0.03103509283944183, "ibn_sina": 0.0744493174585122, "einstein": 0.017683262863843606, "socrates": 0.06570129205667426}, "confidence": 0.56565486816218, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.23681000014767, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "fab59394c58f9dd65f213d2fe89491615f1c92b5ef3dbf6bbfec59942cb9f7ba", "compiled_ids_sha256": "d0e4e9e90aaca1119fab0b84cb552a96f5a742ea58ecad775b431dd1e725d610", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 19.375, 16.125, 17.0, 15.5625, 16.875], "runtime_probabilities": [0.010725441388785839, 0.8004056215286255, 0.03103509359061718, 0.0744493156671524, 0.01768326386809349, 0.06570129841566086], "decide_seconds": 0.16093566499694134, "probabilities_f64_minus_runtime_max_abs": 2.7837776306327555e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013140652062438373, "einstein": 0.010233950116295876, "darwin": 0.8129833552243907, "sagan": 0.04047605788849216, "hypatia": 0.013140652062438373, "socrates": 0.11002533264594462}, "confidence": 0.6083696471512255, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.931314997375, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "3aeab417b11636c4046dba4e7af465fe2b636584ded218286c084cf35db3f09d", "compiled_ids_sha256": "4ed0122584c876807f82edaead790265a969218b57f14fb5de3c5144ecf5106a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 15.125, 19.5, 16.5, 15.375, 17.5], "runtime_probabilities": [0.01314065232872963, 0.010233950801193714, 0.8129833936691284, 0.040476057678461075, 0.01314065232872963, 0.1100253313779831], "decide_seconds": 0.16146684200066375, "probabilities_f64_minus_runtime_max_abs": 3.8444737726628375e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.10049030248432826, "sagan": 0.006034911742395372, "socrates": 0.018588836830659497, "ibn_sina": 0.02879095369969152, "darwin": 0.8413950012521811, "hypatia": 0.0046999939907443325}, "confidence": 0.6604153867919966, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 169.02069299976574, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "f9ff1c68587c98be1aafb24549421ccc60d759a186e65e7d9c32de9712e17b3a", "compiled_ids_sha256": "146c264016969c6ab78aae7142bd6a241931b9b9f80907562acb25accf5a4a89", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 14.5625, 15.6875, 16.125, 19.5, 14.3125], "runtime_probabilities": [0.10049031674861908, 0.006034912075847387, 0.018588837236166, 0.028790956363081932, 0.8413950800895691, 0.004699994809925556], "decide_seconds": 0.1656555789995764, "probabilities_f64_minus_runtime_max_abs": 7.883738795122497e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.026884255997327165, "ibn_sina": 0.02228783104240028, "hypatia": 0.008728040042625774, "einstein": 0.012699223328464383, "sagan": 0.039116361664508915, "darwin": 0.8902842879246734}, "confidence": 0.7158808205711881, "choice": "darwin"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.7014959999069, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "ec8e25a98b9fbce7af2926d67b2dc722125fcc540a8b24ac2e530d7389875c2e", "compiled_ids_sha256": "12537b2200ce45c8f300bfacf497548160c14918c5fd253f2a92610b46f0a5cf", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.875, 15.6875, 14.75, 15.125, 16.25, 19.375], "runtime_probabilities": [0.026884254068136215, 0.022287830710411072, 0.008728040382266045, 0.012699222192168236, 0.03911636024713516, 0.8902842402458191], "decide_seconds": 0.16116681399944355, "probabilities_f64_minus_runtime_max_abs": 4.767885430911889e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:10Z"} {"id": "c-darwin-01-starter-03", "uid": "c016", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "prompt_sha256_by_order": ["e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8"], "request_sha256": "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "compiled_prompt_sha256_by_order": ["48c20dba52dff386d2809ed467785e33fb1eac6e82e19292cd9497781a43ff1b", "d621d8ab1a56a4ed90e1e7646fe59f2f49b01db39f68788e02a52704a33c967a", "46d735edbb9564b6392a1fd9864b5af96d1e82972852bcd9684c302d3b94af77", "7db96ad157113e2bb7ea1485d276629de6e7b9334aceddb0d88b230da827d363", "a8473c49657743c60018490037305ac5afd58b210cdd3e1526fc0e0c5cfd37e6", "5047cc060e275b6fe6eb8ee0793f1860ce1a695c773545ccf05ee1db882b088e"], "compiled_tokens_by_order": [396, 396, 396, 396, 396, 396], "choice": "darwin", "probs": {"darwin": 0.9714386432044175, "einstein": 0.003970047701268099, "hypatia": 0.004226093779639477, "ibn_sina": 0.002728571221370309, "sagan": 0.011487713926557794, "socrates": 0.006148930166746824}, "confidence": 0.9040452334128894, "seconds": 0.17843305133283138, "seconds_all_orders": 1.0705983079969883, "seconds_by_order": [0.1819802840000193, 0.17704377200061572, 0.1768463240005076, 0.17703892199642723, 0.18048574599743006, 0.17720326000198838], "argmax_stable": true, "spread": {"darwin": 0.04983365194930456, "einstein": 0.011030585238343507, "hypatia": 0.004423000816716572, "ibn_sina": 0.007581202973103453, "sagan": 0.03476191809809949, "socrates": 0.011819506600919478}, "spread_2dp": {"darwin": 0.04999999999999993, "einstein": 0.02, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.03, "socrates": 0.01}, "choices_by_order": ["darwin", "darwin", "darwin", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.9714386432044175, "einstein": 0.003970047701268099, "hypatia": 0.004226093779639477, "ibn_sina": 0.002728571221370309, "sagan": 0.011487713926557794, "socrates": 0.006148930166746824}, {"ibn_sina": 0.010309774194473762, "hypatia": 0.003562962622766859, "sagan": 0.029832320794455962, "socrates": 0.013238012106015698, "darwin": 0.9280562973426763, "einstein": 0.015000632939611607}, {"hypatia": 0.007490358841434943, "darwin": 0.921604991255113, "sagan": 0.04049249732777688, "ibn_sina": 0.005833497331195356, "einstein": 0.006610218476813477, "socrates": 0.017968436767666303}, {"ibn_sina": 0.0064209205935084526, "einstein": 0.004413029883886525, "darwin": 0.9529491096293806, "sagan": 0.021053388100072085, "hypatia": 0.003894485203543115, "socrates": 0.011269066589609295}, {"einstein": 0.012131636324427273, "sagan": 0.005730579229677385, "socrates": 0.011396617636469443, "ibn_sina": 0.003938565665819495, "darwin": 0.9637352431188879, "hypatia": 0.0030673580247183715}, {"socrates": 0.009221872733545456, "ibn_sina": 0.004936112772977867, "hypatia": 0.004356104232961225, "einstein": 0.004356104232961225, "sagan": 0.03647321419018159, "darwin": 0.9406565918373727}], "prompt_tokens": 396, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.9714386432044175, "einstein": 0.003970047701268099, "hypatia": 0.004226093779639477, "ibn_sina": 0.002728571221370309, "sagan": 0.011487713926557794, "socrates": 0.006148930166746824}, "confidence": 0.9040452334128894, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 181.95126399950823, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "48c20dba52dff386d2809ed467785e33fb1eac6e82e19292cd9497781a43ff1b", "compiled_ids_sha256": "ec6e841b4247aaa12db66e4d4109704bc2a69c4a9bbb1d5a4325d9879d6dca16", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 13.75, 13.8125, 13.375, 14.8125, 14.1875], "runtime_probabilities": [0.9714386463165283, 0.003970047924667597, 0.004226094111800194, 0.002728571416810155, 0.011487714014947414, 0.006148930173367262], "decide_seconds": 0.17727036100041005, "probabilities_f64_minus_runtime_max_abs": 3.1121107912923662e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.010309774194473762, "hypatia": 0.003562962622766859, "sagan": 0.029832320794455962, "socrates": 0.013238012106015698, "darwin": 0.9280562973426763, "einstein": 0.015000632939611607}, "confidence": 0.7982075063493331, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 177.0179420018394, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "d621d8ab1a56a4ed90e1e7646fe59f2f49b01db39f68788e02a52704a33c967a", "compiled_ids_sha256": "77760dc08b2e87ff16d3e3c31f88c38202a36be0abcd975e12fcfc15ffb768ef", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 13.6875, 15.8125, 15.0, 19.25, 15.125], "runtime_probabilities": [0.010309774428606033, 0.0035629626363515854, 0.029832322150468826, 0.013238011859357357, 0.9280562996864319, 0.015000632964074612], "decide_seconds": 0.1723842470019008, "probabilities_f64_minus_runtime_max_abs": 2.343755634015565e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.007490358841434943, "darwin": 0.921604991255113, "sagan": 0.04049249732777688, "ibn_sina": 0.005833497331195356, "einstein": 0.006610218476813477, "socrates": 0.017968436767666303}, "confidence": 0.789510958647766, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 176.82163399877027, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "46d735edbb9564b6392a1fd9864b5af96d1e82972852bcd9684c302d3b94af77", "compiled_ids_sha256": "107564a7bfa8934860b365ef9a20b4770878cda01b401cc010871b64d7b4e077", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 19.125, 16.0, 14.0625, 14.1875, 15.1875], "runtime_probabilities": [0.007490358781069517, 0.9216048717498779, 0.04049249365925789, 0.00583349633961916, 0.006610217038542032, 0.01796843484044075], "decide_seconds": 0.17225630799657665, "probabilities_f64_minus_runtime_max_abs": 1.1950523504111743e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0064209205935084526, "einstein": 0.004413029883886525, "darwin": 0.9529491096293806, "sagan": 0.021053388100072085, "hypatia": 0.003894485203543115, "socrates": 0.011269066589609295}, "confidence": 0.8572851916452755, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 177.01425199993537, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "7db96ad157113e2bb7ea1485d276629de6e7b9334aceddb0d88b230da827d363", "compiled_ids_sha256": "1362790190d182c95fc794b8ae3ebf66ae9e00a934976662d5b19c9c7c4c9410", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 14.0, 19.375, 15.5625, 13.875, 14.9375], "runtime_probabilities": [0.006420920602977276, 0.004413030110299587, 0.9529491066932678, 0.021053386852145195, 0.0038944853004068136, 0.01126906555145979], "decide_seconds": 0.17234981799992966, "probabilities_f64_minus_runtime_max_abs": 2.9361127973359658e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.012131636324427273, "sagan": 0.005730579229677385, "socrates": 0.011396617636469443, "ibn_sina": 0.003938565665819495, "darwin": 0.9637352431188879, "hypatia": 0.0030673580247183715}, "confidence": 0.883212167820542, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 180.45973599873832, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "a8473c49657743c60018490037305ac5afd58b210cdd3e1526fc0e0c5cfd37e6", "compiled_ids_sha256": "1afcadb575843ea5e49305250c2aeb0f443a6a59cd584fe0bab9d9431f0fc45b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 14.125, 14.8125, 13.75, 19.25, 13.5], "runtime_probabilities": [0.012131636962294579, 0.005730579141527414, 0.011396617628633976, 0.0039385659620165825, 0.9637352824211121, 0.0030673581641167402], "decide_seconds": 0.17590245000246796, "probabilities_f64_minus_runtime_max_abs": 3.930222414183504e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.009221872733545456, "ibn_sina": 0.004936112772977867, "hypatia": 0.004356104232961225, "einstein": 0.004356104232961225, "sagan": 0.03647321419018159, "darwin": 0.9406565918373727}, "confidence": 0.8352965502659062, "choice": "darwin"}}, "usage": {"input_tokens": 396, "output_tokens": 1}, "latency_ms": 177.17725100010284, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 396, "compiled_prompt_sha256": "5047cc060e275b6fe6eb8ee0793f1860ce1a695c773545ccf05ee1db882b088e", "compiled_ids_sha256": "326c215ba64135981dbd53c35d2ce479342b5155494c4d771b2d0043d068ea5e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 14.0, 13.875, 13.875, 16.0, 19.25], "runtime_probabilities": [0.009221873246133327, 0.004936112556606531, 0.004356104414910078, 0.004356104414910078, 0.036473214626312256, 0.9406566023826599], "decide_seconds": 0.17255941700204858, "probabilities_f64_minus_runtime_max_abs": 1.0545287199015263e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:11Z"} {"id": "c-darwin-01-starter-03", "uid": "c017", "task": "kperm", "label": "darwin", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "prompt_sha256_by_order": ["d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169"], "request_sha256": "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "compiled_prompt_sha256_by_order": ["292b2a1853810df91bacbfc1e9d9b69a191c104ded5e174b75e99205cda89199", "13e350e2ee4c4c6afc6af2ffd85fbda214625bb4dc1433ad0c22fae70f15c73a", "391e77e8f3485628fc7e8f6848ca2671886f500c6b41a0cf8eb9eb1fd066b019", "354e5e656727557de97828ddafe9b8576c3a48546f60c17c4401fc471d05e288", "12ea3388b537ba1ca5581a269d64e136849aa4fdf8c9e47fa9538a59c8c34e7c", "5dfff57f55994310a2473a4248ea85d070dd2fa08d515240a3561ad357d461b0"], "compiled_tokens_by_order": [383, 383, 383, 383, 383, 383], "choice": "darwin", "probs": {"darwin": 0.7790394653291113, "einstein": 0.0498023268141071, "hypatia": 0.020760698540417823, "ibn_sina": 0.043950399154955265, "sagan": 0.09304299574117397, "socrates": 0.013404114420234457}, "confidence": 0.5309440264709359, "seconds": 0.17555484149973685, "seconds_all_orders": 1.053329048998421, "seconds_by_order": [0.17955140199774178, 0.17129630600175005, 0.17565504199956194, 0.1746275709992915, 0.17617140899892547, 0.17602731900115032], "argmax_stable": false, "spread": {"darwin": 0.7817518832328645, "einstein": 0.3180021926494261, "hypatia": 0.03112840562614057, "ibn_sina": 0.2952260290849238, "sagan": 0.256848740136354, "socrates": 0.13580774452202807}, "spread_2dp": {"darwin": 0.78, "einstein": 0.31000000000000005, "hypatia": 0.03, "ibn_sina": 0.3, "sagan": 0.25, "socrates": 0.13999999999999999}, "choices_by_order": ["darwin", "darwin", "ibn_sina", "darwin", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.7790394653291113, "einstein": 0.0498023268141071, "hypatia": 0.020760698540417823, "ibn_sina": 0.043950399154955265, "sagan": 0.09304299574117397, "socrates": 0.013404114420234457}, {"ibn_sina": 0.039621888971269746, "hypatia": 0.010664086746370371, "sagan": 0.03496619429169629, "socrates": 0.04489748219537986, "darwin": 0.7958269138988807, "einstein": 0.07402343389640306}, {"hypatia": 0.03693992290067605, "darwin": 0.08861426840748118, "sagan": 0.27295116260326835, "ibn_sina": 0.30929418766668526, "einstein": 0.1461002292109446, "socrates": 0.1461002292109446}, {"ibn_sina": 0.04198559303297238, "einstein": 0.0370521558047755, "darwin": 0.6567651762541227, "sagan": 0.16605624170272543, "hypatia": 0.0370521558047755, "socrates": 0.061088677400628594}, {"einstein": 0.3442849730483174, "sagan": 0.01610242246691435, "socrates": 0.02493991979816393, "ibn_sina": 0.04111897626078811, "darwin": 0.5676299582471812, "hypatia": 0.005923750178634928}, {"socrates": 0.010292484688916526, "ibn_sina": 0.014068158581761485, "hypatia": 0.015941312134421824, "einstein": 0.026282780398891325, "sagan": 0.0630491125556631, "darwin": 0.8703661516403457}], "prompt_tokens": 383, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.7790394653291113, "einstein": 0.0498023268141071, "hypatia": 0.020760698540417823, "ibn_sina": 0.043950399154955265, "sagan": 0.09304299574117397, "socrates": 0.013404114420234457}, "confidence": 0.5309440264709359, "choice": "darwin"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 179.52056200010702, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "292b2a1853810df91bacbfc1e9d9b69a191c104ded5e174b75e99205cda89199", "compiled_ids_sha256": "80f60e0dbaddd871d7e686f5baae1c6e910fcda7924288511f08aec576cdfe5e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 17.0, 16.125, 16.875, 17.625, 15.6875], "runtime_probabilities": [0.7790394425392151, 0.049802325665950775, 0.02076069824397564, 0.0439504012465477, 0.0930429995059967, 0.013404114171862602], "decide_seconds": 0.17496197800210211, "probabilities_f64_minus_runtime_max_abs": 2.278989619242111e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.039621888971269746, "hypatia": 0.010664086746370371, "sagan": 0.03496619429169629, "socrates": 0.04489748219537986, "darwin": 0.7958269138988807, "einstein": 0.07402343389640306}, "confidence": 0.5493907578088213, "choice": "darwin"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 171.26824599836254, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "13e350e2ee4c4c6afc6af2ffd85fbda214625bb4dc1433ad0c22fae70f15c73a", "compiled_ids_sha256": "8f3957dd490479edf1ddf6cd43137e37f5d81c933d3083c3971e4b77a8d53bca", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 15.3125, 16.5, 16.75, 19.625, 17.25], "runtime_probabilities": [0.03962188959121704, 0.01066408772021532, 0.03496619686484337, 0.04489748552441597, 0.7958269119262695, 0.07402343302965164], "decide_seconds": 0.1667810710023332, "probabilities_f64_minus_runtime_max_abs": 3.3290361092785936e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.03693992290067605, "darwin": 0.08861426840748118, "sagan": 0.27295116260326835, "ibn_sina": 0.30929418766668526, "einstein": 0.1461002292109446, "socrates": 0.1461002292109446}, "confidence": 0.09809501223139294, "choice": "ibn_sina"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 175.62778200226603, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "391e77e8f3485628fc7e8f6848ca2671886f500c6b41a0cf8eb9eb1fd066b019", "compiled_ids_sha256": "3b2209be9db2c78c261e60daf8d52264fe6f390eb99a928b67ec300f1e905340", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 17.75, 18.875, 19.0, 18.25, 18.25], "runtime_probabilities": [0.03693992644548416, 0.08861427754163742, 0.2729511559009552, 0.309294193983078, 0.1461002379655838, 0.1461002379655838], "decide_seconds": 0.1711856469992199, "probabilities_f64_minus_runtime_max_abs": 9.134156236934032e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04198559303297238, "einstein": 0.0370521558047755, "darwin": 0.6567651762541227, "sagan": 0.16605624170272543, "hypatia": 0.0370521558047755, "socrates": 0.061088677400628594}, "confidence": 0.3736034441394591, "choice": "darwin"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 174.59978100305307, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "354e5e656727557de97828ddafe9b8576c3a48546f60c17c4401fc471d05e288", "compiled_ids_sha256": "d044de91ccf34555b678168d7bc293e186fa5431d2b77c7f70010d28e63c4a60", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 16.75, 19.625, 18.25, 16.75, 17.25], "runtime_probabilities": [0.041985590010881424, 0.03705215826630592, 0.6567651629447937, 0.16605624556541443, 0.03705215826630592, 0.061088673770427704], "decide_seconds": 0.1700510349983233, "probabilities_f64_minus_runtime_max_abs": 1.330932897225523e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.3442849730483174, "sagan": 0.01610242246691435, "socrates": 0.02493991979816393, "ibn_sina": 0.04111897626078811, "darwin": 0.5676299582471812, "hypatia": 0.005923750178634928}, "confidence": 0.43703682851597814, "choice": "darwin"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 176.14392900213716, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "12ea3388b537ba1ca5581a269d64e136849aa4fdf8c9e47fa9538a59c8c34e7c", "compiled_ids_sha256": "a0b56cf8462290e5e32f8970810e85c61cee48ba7cfdaa90b873eb42285b83b4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.75, 15.6875, 16.125, 16.625, 19.25, 14.6875], "runtime_probabilities": [0.3442849814891815, 0.016102422028779984, 0.024939920753240585, 0.04111897572875023, 0.5676299333572388, 0.005923750344663858], "decide_seconds": 0.1717136120023497, "probabilities_f64_minus_runtime_max_abs": 2.4889942440431412e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.010292484688916526, "ibn_sina": 0.014068158581761485, "hypatia": 0.015941312134421824, "einstein": 0.026282780398891325, "sagan": 0.0630491125556631, "darwin": 0.8703661516403457}, "confidence": 0.6853348583057075, "choice": "darwin"}}, "usage": {"input_tokens": 383, "output_tokens": 1}, "latency_ms": 175.9997290027968, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 383, "compiled_prompt_sha256": "5dfff57f55994310a2473a4248ea85d070dd2fa08d515240a3561ad357d461b0", "compiled_ids_sha256": "6e87572a499b5ef32fba05091c0ae7c63d9cd0c298c0a8266754befaf9c58f1c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 15.5, 15.625, 16.125, 17.0, 19.625], "runtime_probabilities": [0.010292484425008297, 0.014068158343434334, 0.015941312536597252, 0.02628277987241745, 0.06304911524057388, 0.8703661561012268], "decide_seconds": 0.17149099499874865, "probabilities_f64_minus_runtime_max_abs": 4.46088110894749e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:12Z"} {"id": "c-einstein-01-dessert-01", "uid": "c018", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "prompt_sha256_by_order": ["f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3"], "request_sha256": "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "compiled_prompt_sha256_by_order": ["511f397ccbe1d23a7056a582c58d20ec2dc37e7b408ccc37042cd3604b440fdb", "16420fcd2d02a093c6ffb478645762f51ec9e0a475f29d62048c3b5eaea2a2ab", "399555e241d43dd37e0db5aa8d6ff1a45f2de6766a64bb3c013770f4ae212881", "8ceabf110cc29e0400fc03f2bb9ab29ab7a8a40842ad988676de7d57a6b7af10", "8f6060573b36acad3b9b762461c3cf0d95ab019b9cb99e5b823a2cd0ff0802e8", "ef3fabe0a28442c41a3c1edd976bdc9de84545f815124210cbb69daad8e48260"], "compiled_tokens_by_order": [373, 373, 373, 373, 373, 373], "choice": "sagan", "probs": {"darwin": 0.0020239319793062278, "einstein": 0.008004806260128634, "hypatia": 0.0016778985440613545, "ibn_sina": 0.0007925831511074295, "sagan": 0.9849019999623201, "socrates": 0.0025987801030763246}, "confidence": 0.9452865403262539, "seconds": 0.1738703095003681, "seconds_all_orders": 1.0432218570022087, "seconds_by_order": [0.17244252700038487, 0.1711518869997235, 0.17389119600193226, 0.17113183699984802, 0.17932225500044296, 0.1752821549998771], "argmax_stable": true, "spread": {"darwin": 0.00237394236579681, "einstein": 0.005387146068786949, "hypatia": 0.00135789071076194, "ibn_sina": 0.001229621832018214, "sagan": 0.005994015373945283, "socrates": 0.001745496719402518}, "spread_2dp": {"darwin": 0.0, "einstein": 0.01, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.010000000000000009, "socrates": 0.0}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0020239319793062278, "einstein": 0.008004806260128634, "hypatia": 0.0016778985440613545, "ibn_sina": 0.0007925831511074295, "sagan": 0.9849019999623201, "socrates": 0.0025987801030763246}, {"ibn_sina": 0.0013884861622114094, "hypatia": 0.0010813541104140554, "sagan": 0.983103162975889, "socrates": 0.0025940336595020467, "darwin": 0.0015733609468145532, "einstein": 0.010259602145169026}, {"hypatia": 0.0014870474635946688, "darwin": 0.0009601100035982323, "sagan": 0.9890971783498342, "ibn_sina": 0.001232805647436289, "einstein": 0.005190305641969146, "socrates": 0.0020325528935674526}, {"ibn_sina": 0.0020222049831256435, "einstein": 0.005851444137456287, "darwin": 0.003334052369395042, "sagan": 0.9840615952404341, "hypatia": 0.0024392448211759954, "socrates": 0.002291458448412851}, {"einstein": 0.007998131093420936, "sagan": 0.9840806952577654, "socrates": 0.0037780496129699707, "ibn_sina": 0.0010168482586485738, "darwin": 0.001899722648542193, "hypatia": 0.001226553128652959}, {"socrates": 0.0023015852801826116, "ibn_sina": 0.0010213222770032425, "hypatia": 0.00190808121524678, "einstein": 0.004872456076382077, "sagan": 0.9884105400065872, "darwin": 0.0014860151445980302}], "prompt_tokens": 373, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0020239319793062278, "einstein": 0.008004806260128634, "hypatia": 0.0016778985440613545, "ibn_sina": 0.0007925831511074295, "sagan": 0.9849019999623201, "socrates": 0.0025987801030763246}, "confidence": 0.9452865403262539, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 172.41315699720872, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "511f397ccbe1d23a7056a582c58d20ec2dc37e7b408ccc37042cd3604b440fdb", "compiled_ids_sha256": "95907c564f0990963bab66233f57176e924cae32017e3e77c3639ed1e429d350", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 14.6875, 13.125, 12.375, 19.5, 13.5625], "runtime_probabilities": [0.002023932058364153, 0.008004807867109776, 0.0016778988065198064, 0.000792583217844367, 0.9849020838737488, 0.0025987804401665926], "decide_seconds": 0.1679164019988093, "probabilities_f64_minus_runtime_max_abs": 8.391142869434987e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0013884861622114094, "hypatia": 0.0010813541104140554, "sagan": 0.983103162975889, "socrates": 0.0025940336595020467, "darwin": 0.0015733609468145532, "einstein": 0.010259602145169026}, "confidence": 0.9409184630326485, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 171.12707699925522, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "16420fcd2d02a093c6ffb478645762f51ec9e0a475f29d62048c3b5eaea2a2ab", "compiled_ids_sha256": "5a60a66c897a80858b5972b6551068a349f2461dd904e13efc7f8ed7d6b5ae99", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 12.9375, 19.75, 13.8125, 13.3125, 15.1875], "runtime_probabilities": [0.0013884862419217825, 0.0010813541011884809, 0.9831031560897827, 0.002594033721834421, 0.0015733609907329082, 0.010259603150188923], "decide_seconds": 0.1666753309982596, "probabilities_f64_minus_runtime_max_abs": 6.8861062407776785e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0014870474635946688, "darwin": 0.0009601100035982323, "sagan": 0.9890971783498342, "ibn_sina": 0.001232805647436289, "einstein": 0.005190305641969146, "socrates": 0.0020325528935674526}, "confidence": 0.9579412419135643, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 173.86645600345219, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "399555e241d43dd37e0db5aa8d6ff1a45f2de6766a64bb3c013770f4ae212881", "compiled_ids_sha256": "5a4e05941fdf182cba2855a4f92d8814a8b5a12c095150aa66d75e2d96a6935f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 12.9375, 19.875, 13.1875, 14.625, 13.6875], "runtime_probabilities": [0.0014870474115014076, 0.0009601099300198257, 0.9890971183776855, 0.0012328056618571281, 0.005190305411815643, 0.002032552845776081], "decide_seconds": 0.1695827690018632, "probabilities_f64_minus_runtime_max_abs": 5.99721486915783e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0020222049831256435, "einstein": 0.005851444137456287, "darwin": 0.003334052369395042, "sagan": 0.9840615952404341, "hypatia": 0.0024392448211759954, "socrates": 0.002291458448412851}, "confidence": 0.9408080760389891, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 171.1066980024043, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "8ceabf110cc29e0400fc03f2bb9ab29ab7a8a40842ad988676de7d57a6b7af10", "compiled_ids_sha256": "a9c3f5411aea2f5cecb8a7fc2b034aaabeca50234ed8c268fc16da47b71328ab", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.4375, 14.5, 13.9375, 19.625, 13.625, 13.5625], "runtime_probabilities": [0.0020222049206495285, 0.005851443391293287, 0.003334052162244916, 0.9840615391731262, 0.002439244883134961, 0.0022914581932127476], "decide_seconds": 0.166765440000745, "probabilities_f64_minus_runtime_max_abs": 5.606730790752579e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.007998131093420936, "sagan": 0.9840806952577654, "socrates": 0.0037780496129699707, "ibn_sina": 0.0010168482586485738, "darwin": 0.001899722648542193, "hypatia": 0.001226553128652959}, "confidence": 0.9427264323601221, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 179.29777500103228, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "8f6060573b36acad3b9b762461c3cf0d95ab019b9cb99e5b823a2cd0ff0802e8", "compiled_ids_sha256": "3dbef42a3452a01e857cdda10efc3e47e726fe41a45b3674dbc090ab674062a4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 19.625, 14.0625, 12.75, 13.375, 12.9375], "runtime_probabilities": [0.007998131215572357, 0.9840806126594543, 0.0037780492566525936, 0.0010168481385335326, 0.0018997224979102612, 0.0012265531113371253], "decide_seconds": 0.17500545700022485, "probabilities_f64_minus_runtime_max_abs": 8.259831107970683e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0023015852801826116, "ibn_sina": 0.0010213222770032425, "hypatia": 0.00190808121524678, "einstein": 0.004872456076382077, "sagan": 0.9884105400065872, "darwin": 0.0014860151445980302}, "confidence": 0.9552944337947654, "choice": "sagan"}}, "usage": {"input_tokens": 373, "output_tokens": 1}, "latency_ms": 175.2576049984782, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 373, "compiled_prompt_sha256": "ef3fabe0a28442c41a3c1edd976bdc9de84545f815124210cbb69daad8e48260", "compiled_ids_sha256": "af18a24b0165ecff509fd261726553ffacba047d434f04cc37b16bc02e0570d3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.6875, 12.875, 13.5, 14.4375, 19.75, 13.25], "runtime_probabilities": [0.0023015853948891163, 0.0010213223285973072, 0.0019080813508480787, 0.004872456192970276, 0.9884105920791626, 0.0014860151568427682], "decide_seconds": 0.17094494899720303, "probabilities_f64_minus_runtime_max_abs": 5.2072575362771545e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:13Z"} {"id": "c-einstein-01-dessert-01", "uid": "c019", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "prompt_sha256_by_order": ["b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd"], "request_sha256": "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "compiled_prompt_sha256_by_order": ["482a99aea3518461af305b2fde5179c25aa0874e22f2f00e33c301df075be35a", "e8f61a055e57f14754c74acd7cc3bb1a481e3083be49c4f112106963b4d19de2", "83792c4ef43956e9c0615090fdc2a9d10d767913a586f733583dd6c772ddac9f", "fc6239bffaf1148332bf71061007eaa3073fe882d2532f6871925929decbd17b", "6d488638770e470d6e8b24ecb6a22f3ac30f19d16d04d3d47158a932c3d94110", "b0631ff6dd0c6bb01252176fe7b6fb465b6fddfad854d371cab3da4586843feb"], "compiled_tokens_by_order": [380, 380, 380, 380, 380, 380], "choice": "sagan", "probs": {"darwin": 0.0020390750739017844, "einstein": 0.0016904526104204313, "hypatia": 0.0020390750739017844, "ibn_sina": 0.0007985132721563486, "sagan": 0.9922710540141736, "socrates": 0.0011618299554461745}, "confidence": 0.9680197488751763, "seconds": 0.17417148400030177, "seconds_all_orders": 1.0450289040018106, "seconds_by_order": [0.17160707299990463, 0.17175966200011317, 0.17928838499938138, 0.17752881900014472, 0.17148875500060967, 0.17335621000165702], "argmax_stable": true, "spread": {"darwin": 0.013792193457658074, "einstein": 0.008544556165280654, "hypatia": 0.0049903344744731265, "ibn_sina": 0.001776305415123412, "sagan": 0.016450101852715338, "socrates": 0.0046055387996126376}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.01, "ibn_sina": 0.0, "sagan": 0.010000000000000009, "socrates": 0.01}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0020390750739017844, "einstein": 0.0016904526104204313, "hypatia": 0.0020390750739017844, "ibn_sina": 0.0007985132721563486, "sagan": 0.9922710540141736, "socrates": 0.0011618299554461745}, {"ibn_sina": 0.0020153927237529278, "hypatia": 0.0012223974783179044, "sagan": 0.9807465589896854, "socrates": 0.0020153927237529278, "darwin": 0.0037652493087897234, "einstein": 0.010235008775701085}, {"hypatia": 0.0016897168081777877, "darwin": 0.001024865050391599, "sagan": 0.9918391488176876, "ibn_sina": 0.001236223225356704, "einstein": 0.0033604031294061604, "socrates": 0.0008496429689800588}, {"ibn_sina": 0.0025748186872797604, "einstein": 0.0025748186872797604, "darwin": 0.014817058508049673, "sagan": 0.9758209521614583, "hypatia": 0.002917651812418617, "socrates": 0.0012947001435139487}, {"einstein": 0.007937242638484108, "sagan": 0.9765890509764291, "socrates": 0.005455181768592696, "ibn_sina": 0.001562935744446063, "darwin": 0.006181530782279168, "hypatia": 0.002274058089768929}, {"socrates": 0.0031239578113323994, "ibn_sina": 0.0014756531822474506, "hypatia": 0.006212731952791031, "einstein": 0.004545331795118107, "sagan": 0.9815183674471787, "darwin": 0.0031239578113323994}], "prompt_tokens": 380, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0020390750739017844, "einstein": 0.0016904526104204313, "hypatia": 0.0020390750739017844, "ibn_sina": 0.0007985132721563486, "sagan": 0.9922710540141736, "socrates": 0.0011618299554461745}, "confidence": 0.9680197488751763, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.57868399954168, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "482a99aea3518461af305b2fde5179c25aa0874e22f2f00e33c301df075be35a", "compiled_ids_sha256": "0d8c04b699286490547a95aa4750df3f99b6a6ff4c2170fbcda9228d7de90d1c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 13.125, 13.3125, 12.375, 19.5, 12.75], "runtime_probabilities": [0.002039075130596757, 0.0016904528019949794, 0.002039075130596757, 0.0007985132979229093, 0.9922710657119751, 0.001161829917691648], "decide_seconds": 0.16696357900218572, "probabilities_f64_minus_runtime_max_abs": 1.1697801505228256e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0020153927237529278, "hypatia": 0.0012223974783179044, "sagan": 0.9807465589896854, "socrates": 0.0020153927237529278, "darwin": 0.0037652493087897234, "einstein": 0.010235008775701085}, "confidence": 0.9329162761939873, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.73411199837574, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "e8f61a055e57f14754c74acd7cc3bb1a481e3083be49c4f112106963b4d19de2", "compiled_ids_sha256": "bb20ea2c7640a34981ca59f525ad9ab43740f166487587c88a26856b242b01e4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 13.0625, 19.75, 13.5625, 14.1875, 15.1875], "runtime_probabilities": [0.00201539252884686, 0.0012223974335938692, 0.9807464480400085, 0.00201539252884686, 0.0037652491591870785, 0.010235008783638477], "decide_seconds": 0.16736966600001324, "probabilities_f64_minus_runtime_max_abs": 1.1094967689029289e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0016897168081777877, "darwin": 0.001024865050391599, "sagan": 0.9918391488176876, "ibn_sina": 0.001236223225356704, "einstein": 0.0033604031294061604, "socrates": 0.0008496429689800588}, "confidence": 0.9668524904184518, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 179.26336500022444, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "83792c4ef43956e9c0615090fdc2a9d10d767913a586f733583dd6c772ddac9f", "compiled_ids_sha256": "fb4d09450824be56ff3abbd515a120ee9004d99464b5047a0b02f36efeeaf1a2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 13.0, 19.875, 13.1875, 14.1875, 12.8125], "runtime_probabilities": [0.0016897170571610332, 0.0010248650796711445, 0.9918392300605774, 0.0012362233828753233, 0.0033604034688323736, 0.0008496430236846209], "decide_seconds": 0.17473906900340808, "probabilities_f64_minus_runtime_max_abs": 8.124288974897809e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0025748186872797604, "einstein": 0.0025748186872797604, "darwin": 0.014817058508049673, "sagan": 0.9758209521614583, "hypatia": 0.002917651812418617, "socrates": 0.0012947001435139487}, "confidence": 0.9203940001286905, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 177.50315900048008, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "fc6239bffaf1148332bf71061007eaa3073fe882d2532f6871925929decbd17b", "compiled_ids_sha256": "42c5db44500bc5ba0789dc5d445dbc361bd1007559da29ad18c0e76fc6cb0d29", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.6875, 13.6875, 15.4375, 19.625, 13.8125, 13.0], "runtime_probabilities": [0.0025748189073055983, 0.0025748189073055983, 0.01481705904006958, 0.9758209586143494, 0.002917652018368244, 0.0012947001960128546], "decide_seconds": 0.17310259199803113, "probabilities_f64_minus_runtime_max_abs": 6.452891110342307e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.007937242638484108, "sagan": 0.9765890509764291, "socrates": 0.005455181768592696, "ibn_sina": 0.001562935744446063, "darwin": 0.006181530782279168, "hypatia": 0.002274058089768929}, "confidence": 0.9188909267596503, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.46397500255262, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "6d488638770e470d6e8b24ecb6a22f3ac30f19d16d04d3d47158a932c3d94110", "compiled_ids_sha256": "93e3ae75a4cdf7eff2a94de5cc064b009cc7556fe7bd27a990a72760a028903a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 19.625, 14.4375, 13.1875, 14.5625, 13.5625], "runtime_probabilities": [0.007937244139611721, 0.9765891432762146, 0.0054551828652620316, 0.0015629359986633062, 0.006181531585752964, 0.0022740582935512066], "decide_seconds": 0.16708221799854073, "probabilities_f64_minus_runtime_max_abs": 9.229978548130902e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0031239578113323994, "ibn_sina": 0.0014756531822474506, "hypatia": 0.006212731952791031, "einstein": 0.004545331795118107, "sagan": 0.9815183674471787, "darwin": 0.0031239578113323994}, "confidence": 0.9329960699091825, "choice": "sagan"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 173.33060999953886, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "b0631ff6dd0c6bb01252176fe7b6fb465b6fddfad854d371cab3da4586843feb", "compiled_ids_sha256": "d352072ebd66245e2636822089274d484f0734f5444f780573b3d0a52095ed2f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.875, 13.125, 14.5625, 14.25, 19.625, 13.875], "runtime_probabilities": [0.0031239581294357777, 0.0014756532618775964, 0.0062127322889864445, 0.0045453323982656, 0.9815183877944946, 0.0031239581294357777], "decide_seconds": 0.1688912049976352, "probabilities_f64_minus_runtime_max_abs": 2.034731594324768e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:14Z"} {"id": "c-einstein-01-dessert-02", "uid": "c020", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "prompt_sha256_by_order": ["ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf"], "request_sha256": "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "compiled_prompt_sha256_by_order": ["bf3f0bcb078d10695d6d82ba348cba3a8f6b4084459aec08e47402eef4080ca8", "d9c4f2b5904f1b737efd6cbb3b7d22702446d54d9ec55b979db7e39cbf093b36", "f3088cabb7dce69dfba27a2fb40b8b3a2bdcbefc96419894e3a8f4e9f2799ebb", "ad594906545a63fa7e9cb9afe30af90b8eb6cf2cd1d0752ba630fe842f9d2242", "f1b6ab5cfdb6fb47065c23daf89a5b247c20aeec581fc69eadb52d837b09a5d2", "a1323d1db85f8a1781b2f63f6b4cc3844dfd90bd9b75afb4402d165fe4eced3f"], "compiled_tokens_by_order": [349, 349, 349, 349, 349, 349], "choice": "sagan", "probs": {"darwin": 0.012850766802806201, "einstein": 0.018697755369343615, "hypatia": 0.06947058898486998, "ibn_sina": 0.005032432587090327, "sagan": 0.7468792182208278, "socrates": 0.14706923803506203}, "confidence": 0.5299880663913374, "seconds": 0.1721412211675973, "seconds_all_orders": 1.0328473270055838, "seconds_by_order": [0.17122138700142386, 0.1707351800032484, 0.17121275599856745, 0.17229089900138206, 0.17136375600239262, 0.17602334899856942], "argmax_stable": true, "spread": {"darwin": 0.04565497579386124, "einstein": 0.12848397939634956, "hypatia": 0.13105773056827563, "ibn_sina": 0.010422947729937244, "sagan": 0.3565429372390678, "socrates": 0.19045701567956133}, "spread_2dp": {"darwin": 0.049999999999999996, "einstein": 0.13, "hypatia": 0.13, "ibn_sina": 0.02, "sagan": 0.36, "socrates": 0.19}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.012850766802806201, "einstein": 0.018697755369343615, "hypatia": 0.06947058898486998, "ibn_sina": 0.005032432587090327, "sagan": 0.7468792182208278, "socrates": 0.14706923803506203}, {"ibn_sina": 0.01504155358292646, "hypatia": 0.028101321674802518, "sagan": 0.724742638115542, "socrates": 0.059490498452395055, "darwin": 0.029913701211095624, "einstein": 0.14271028696323837}, {"hypatia": 0.018958342931573006, "darwin": 0.00953285349199709, "sagan": 0.7114065700720555, "ibn_sina": 0.010802138187769197, "einstein": 0.04547870047310625, "socrates": 0.20382139484349887}, {"ibn_sina": 0.010048749859191866, "einstein": 0.014620844772770024, "darwin": 0.024105697772470718, "sagan": 0.7044717524473674, "hypatia": 0.10803424223975468, "socrates": 0.1387187129084454}, {"einstein": 0.09098934802736211, "sagan": 0.5236075456462747, "socrates": 0.1699904816215128, "ibn_sina": 0.010208721919143399, "darwin": 0.05518782928585833, "hypatia": 0.15001607349984863}, {"socrates": 0.013364379163937548, "ibn_sina": 0.0046186058529892144, "hypatia": 0.043820112265421006, "einstein": 0.014226307566888812, "sagan": 0.8801504828853425, "darwin": 0.043820112265421006}], "prompt_tokens": 349, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.012850766802806201, "einstein": 0.018697755369343615, "hypatia": 0.06947058898486998, "ibn_sina": 0.005032432587090327, "sagan": 0.7468792182208278, "socrates": 0.14706923803506203}, "confidence": 0.5299880663913374, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.19150700091268, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "bf3f0bcb078d10695d6d82ba348cba3a8f6b4084459aec08e47402eef4080ca8", "compiled_ids_sha256": "2f7bcc4bbd9219b9afcae08151a8d821af356541e49f69a31dd887caf3fb2636", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 15.1875, 16.5, 13.875, 18.875, 17.25], "runtime_probabilities": [0.012850766070187092, 0.01869775541126728, 0.06947058439254761, 0.0050324322655797005, 0.7468792200088501, 0.14706923067569733], "decide_seconds": 0.16678059999685502, "probabilities_f64_minus_runtime_max_abs": 7.359364700976201e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01504155358292646, "hypatia": 0.028101321674802518, "sagan": 0.724742638115542, "socrates": 0.059490498452395055, "darwin": 0.029913701211095624, "einstein": 0.14271028696323837}, "confidence": 0.4711709992525701, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 170.7109599992691, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "d9c4f2b5904f1b737efd6cbb3b7d22702446d54d9ec55b979db7e39cbf093b36", "compiled_ids_sha256": "92ea54ebbfdf6268cbbad32eb069488c35be4d39a71981556e1f9fd8e8069bf8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 15.625, 18.875, 16.375, 15.6875, 17.25], "runtime_probabilities": [0.015041554346680641, 0.028101323172450066, 0.7247426509857178, 0.05949050188064575, 0.029913704842329025, 0.14271028339862823], "decide_seconds": 0.1666665809971164, "probabilities_f64_minus_runtime_max_abs": 1.287017581397265e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.018958342931573006, "darwin": 0.00953285349199709, "sagan": 0.7114065700720555, "ibn_sina": 0.010802138187769197, "einstein": 0.04547870047310625, "socrates": 0.20382139484349887}, "confidence": 0.5114172708664214, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.18847600067966, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "f3088cabb7dce69dfba27a2fb40b8b3a2bdcbefc96419894e3a8f4e9f2799ebb", "compiled_ids_sha256": "a08669298e97b3920d7fa38969239585ad20e8508a52b90a80fb189b725013da", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 14.8125, 19.125, 14.9375, 16.375, 17.875], "runtime_probabilities": [0.01895834133028984, 0.009532853029668331, 0.7114065289497375, 0.010802137665450573, 0.0454786978662014, 0.20382139086723328], "decide_seconds": 0.16705564899893943, "probabilities_f64_minus_runtime_max_abs": 4.1122317995245794e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.010048749859191866, "einstein": 0.014620844772770024, "darwin": 0.024105697772470718, "sagan": 0.7044717524473674, "hypatia": 0.10803424223975468, "socrates": 0.1387187129084454}, "confidence": 0.46476697131581046, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 172.2666890018445, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "ad594906545a63fa7e9cb9afe30af90b8eb6cf2cd1d0752ba630fe842f9d2242", "compiled_ids_sha256": "d3f9b8e470ecce0a23441b6e2fa827a793bf0b5860fc115063ddf63fedc57dad", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 15.125, 15.625, 19.0, 17.125, 17.375], "runtime_probabilities": [0.01004874985665083, 0.014620844274759293, 0.024105697870254517, 0.7044717669487, 0.10803424566984177, 0.13871870934963226], "decide_seconds": 0.1681309199993848, "probabilities_f64_minus_runtime_max_abs": 1.450133257918651e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.09098934802736211, "sagan": 0.5236075456462747, "socrates": 0.1699904816215128, "ibn_sina": 0.010208721919143399, "darwin": 0.05518782928585833, "hypatia": 0.15001607349984863}, "confidence": 0.24690056989672549, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.33947600086685, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "f1b6ab5cfdb6fb47065c23daf89a5b247c20aeec581fc69eadb52d837b09a5d2", "compiled_ids_sha256": "81ba9aab8a5579f40d1acd3a4a3eaed74f975055ece0a5a33d5968ccf2097c5a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 18.5, 17.375, 14.5625, 16.25, 17.25], "runtime_probabilities": [0.09098934382200241, 0.5236075520515442, 0.16999049484729767, 0.01020872313529253, 0.055187828838825226, 0.15001608431339264], "decide_seconds": 0.16723805699803052, "probabilities_f64_minus_runtime_max_abs": 1.322578485618564e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.013364379163937548, "ibn_sina": 0.0046186058529892144, "hypatia": 0.043820112265421006, "einstein": 0.014226307566888812, "sagan": 0.8801504828853425, "darwin": 0.043820112265421006}, "confidence": 0.704492928472302, "choice": "sagan"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 175.99900999994134, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "a1323d1db85f8a1781b2f63f6b4cc3844dfd90bd9b75afb4402d165fe4eced3f", "compiled_ids_sha256": "51d5b5d24649adc2d94c619bd69fcb9c6532e86126594aed858286bb4bd16709", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 13.875, 16.125, 15.0, 19.125, 16.125], "runtime_probabilities": [0.013364379294216633, 0.004618606064468622, 0.04382011294364929, 0.014226308092474937, 0.8801504969596863, 0.04382011294364929], "decide_seconds": 0.17186506199868745, "probabilities_f64_minus_runtime_max_abs": 1.4074343801517841e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:15Z"} {"id": "c-einstein-01-dessert-02", "uid": "c021", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "prompt_sha256_by_order": ["dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516"], "request_sha256": "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "compiled_prompt_sha256_by_order": ["dc15cfd6dc8dccc74630a73d746dd43d7892002655d25aa889cf89abbcc1d5b6", "fec807287013fcce68866c0c36f89b0670fe2f90ae7c84a031155b0d16ad2e3e", "0434d17312dd3dd6947c54852d9e5ec5618e9f1160d8679415f29bf58ad9862d", "185f4da2619e40cd99bea0f30edf8aca1312da51d8c3ddf78bb79af6dee3dd8c", "04b7de9cd69e0223c7e901130cc94ae88621b1a4f60101fc891b6dfedd8ec3b2", "06141ca6839adc7b1fbce8f003e7b93e9e93f83ae42f445551958d390945de70"], "compiled_tokens_by_order": [287, 287, 287, 287, 287, 287], "choice": "sagan", "probs": {"darwin": 0.01230477217360307, "einstein": 0.010858923330213828, "hypatia": 0.13228876789007202, "ibn_sina": 0.2800553238209527, "sagan": 0.3173442569608417, "socrates": 0.24714795582431665}, "confidence": 0.19802066791046835, "seconds": 0.16687151800033462, "seconds_all_orders": 1.0012291080020077, "seconds_by_order": [0.16535893100081012, 0.1690587230004894, 0.167413975999807, 0.1695788690012705, 0.16494022400001995, 0.16487838499961072], "argmax_stable": false, "spread": {"darwin": 0.11792855500527308, "einstein": 0.041508704738808334, "hypatia": 0.26143925256121914, "ibn_sina": 0.5960466885360964, "sagan": 0.5955600336831813, "socrates": 0.22643763136349232}, "spread_2dp": {"darwin": 0.12000000000000001, "einstein": 0.04, "hypatia": 0.26, "ibn_sina": 0.59, "sagan": 0.59, "socrates": 0.23}, "choices_by_order": ["sagan", "ibn_sina", "ibn_sina", "sagan", "ibn_sina", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01230477217360307, "einstein": 0.010858923330213828, "hypatia": 0.13228876789007202, "ibn_sina": 0.2800553238209527, "sagan": 0.3173442569608417, "socrates": 0.24714795582431665}, {"ibn_sina": 0.722912806627186, "hypatia": 0.08633962229350993, "sagan": 0.07619444924434639, "socrates": 0.04621426956661418, "darwin": 0.015971224199321364, "einstein": 0.05236762806902216}, {"hypatia": 0.01892082120115055, "darwin": 0.00893756308581659, "sagan": 0.05143212445061018, "ibn_sina": 0.7099985775254837, "einstein": 0.03119516037345086, "socrates": 0.17951575336348824}, {"ibn_sina": 0.28562826122435503, "einstein": 0.011789286312623388, "darwin": 0.0171533103693429, "sagan": 0.4155866678537614, "hypatia": 0.13492123711995863, "socrates": 0.13492123711995863}, {"einstein": 0.017922819883383064, "sagan": 0.04871927560374466, "socrates": 0.10313870726248492, "ibn_sina": 0.5237815744319468, "darwin": 0.026077549056070758, "hypatia": 0.2803600737623697}, {"socrates": 0.020710324460824332, "ibn_sina": 0.12686611809108966, "hypatia": 0.06790653960009556, "einstein": 0.01337159046997476, "sagan": 0.644279309286926, "darwin": 0.12686611809108966}], "prompt_tokens": 287, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01230477217360307, "einstein": 0.010858923330213828, "hypatia": 0.13228876789007202, "ibn_sina": 0.2800553238209527, "sagan": 0.3173442569608417, "socrates": 0.24714795582431665}, "confidence": 0.19802066791046835, "choice": "sagan"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 165.33181100021466, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "dc15cfd6dc8dccc74630a73d746dd43d7892002655d25aa889cf89abbcc1d5b6", "compiled_ids_sha256": "770cba00074ef91bbf9b863248bc0cf872ece1572ae250a4ff112e7036365292", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.875, 17.375, 18.125, 18.25, 18.0], "runtime_probabilities": [0.012304771691560745, 0.010858923196792603, 0.13228875398635864, 0.28005531430244446, 0.3173442482948303, 0.24714794754981995], "decide_seconds": 0.16161192000072333, "probabilities_f64_minus_runtime_max_abs": 1.390371337506302e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.722912806627186, "hypatia": 0.08633962229350993, "sagan": 0.07619444924434639, "socrates": 0.04621426956661418, "darwin": 0.015971224199321364, "einstein": 0.05236762806902216}, "confidence": 0.43919904083950845, "choice": "ibn_sina"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 169.035802999133, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "fec807287013fcce68866c0c36f89b0670fe2f90ae7c84a031155b0d16ad2e3e", "compiled_ids_sha256": "6a533cf70bffa70da3b7dac36b88a0245aa638de7e2abcf91ebd7100ef1c6388", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.0, 16.875, 16.75, 16.25, 15.1875, 16.375], "runtime_probabilities": [0.7229127883911133, 0.08633962273597717, 0.07619445025920868, 0.0462142676115036, 0.015971224755048752, 0.05236763134598732], "decide_seconds": 0.1655489500008116, "probabilities_f64_minus_runtime_max_abs": 1.8236072762967126e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.01892082120115055, "darwin": 0.00893756308581659, "sagan": 0.05143212445061018, "ibn_sina": 0.7099985775254837, "einstein": 0.03119516037345086, "socrates": 0.17951575336348824}, "confidence": 0.48123031719554454, "choice": "ibn_sina"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 167.38987600183464, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "0434d17312dd3dd6947c54852d9e5ec5618e9f1160d8679415f29bf58ad9862d", "compiled_ids_sha256": "9b367ebc79e571e8515edfe2e2a9b6c5683bc15bdd516765585784cd0351232d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 14.75, 16.5, 19.125, 16.0, 17.75], "runtime_probabilities": [0.01892082206904888, 0.008937563747167587, 0.05143212899565697, 0.709998607635498, 0.03119516186416149, 0.1795157641172409], "decide_seconds": 0.16381578300206456, "probabilities_f64_minus_runtime_max_abs": 3.0110014370698934e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.28562826122435503, "einstein": 0.011789286312623388, "darwin": 0.0171533103693429, "sagan": 0.4155866678537614, "hypatia": 0.13492123711995863, "socrates": 0.13492123711995863}, "confidence": 0.22678070683647433, "choice": "sagan"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 169.556000000739, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "185f4da2619e40cd99bea0f30edf8aca1312da51d8c3ddf78bb79af6dee3dd8c", "compiled_ids_sha256": "e63e6e5f27f1b9fc3d165a78ab8f5d3391828c2d5638850d39c7d51423bcd196", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.125, 14.9375, 15.3125, 18.5, 17.375, 17.375], "runtime_probabilities": [0.2856282591819763, 0.011789285577833652, 0.017153309658169746, 0.4155866503715515, 0.13492122292518616, 0.13492122292518616], "decide_seconds": 0.16609060599876102, "probabilities_f64_minus_runtime_max_abs": 1.748220990194227e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.017922819883383064, "sagan": 0.04871927560374466, "socrates": 0.10313870726248492, "ibn_sina": 0.5237815744319468, "darwin": 0.026077549056070758, "hypatia": 0.2803600737623697}, "confidence": 0.30574586887839816, "choice": "ibn_sina"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 164.91810500156134, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "04b7de9cd69e0223c7e901130cc94ae88621b1a4f60101fc891b6dfedd8ec3b2", "compiled_ids_sha256": "923d2ecf1e4b42fbbc99c73a2362495010208919cdf1f205bf598ce4d5b2d773", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 16.375, 17.125, 18.75, 15.75, 18.125], "runtime_probabilities": [0.017922820523381233, 0.04871927574276924, 0.10313870757818222, 0.5237815976142883, 0.026077549904584885, 0.2803601026535034], "decide_seconds": 0.16144225199968787, "probabilities_f64_minus_runtime_max_abs": 2.889113370763141e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.020710324460824332, "ibn_sina": 0.12686611809108966, "hypatia": 0.06790653960009556, "einstein": 0.01337159046997476, "sagan": 0.644279309286926, "darwin": 0.12686611809108966}, "confidence": 0.37059956567870134, "choice": "sagan"}}, "usage": {"input_tokens": 287, "output_tokens": 1}, "latency_ms": 164.85657500015805, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 287, "compiled_prompt_sha256": "06141ca6839adc7b1fbce8f003e7b93e9e93f83ae42f445551958d390945de70", "compiled_ids_sha256": "7a177e87768ffdb513b8c2dba2029ad41206a787366d8badcf71396f373a39c8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 17.25, 16.625, 15.0, 18.875, 17.25], "runtime_probabilities": [0.020710324868559837, 0.12686610221862793, 0.06790653616189957, 0.013371589593589306, 0.6442793011665344, 0.12686610221862793], "decide_seconds": 0.1613197830010904, "probabilities_f64_minus_runtime_max_abs": 1.587246173384571e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:16Z"} {"id": "c-einstein-01-dessert-03", "uid": "c022", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "prompt_sha256_by_order": ["78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c"], "request_sha256": "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "compiled_prompt_sha256_by_order": ["7f48d90264fa637fb31af7ae8698c998de26360d2468e492bee50203c0985399", "8d80177bfd46c0d4a66807221004faa3440fa2c323586ff0418a3061b0726218", "88d56b8abe9c5079891b409265cd812ec9c063b6e44d366648d272241bca7f80", "c52b8e78a74d935c029bc1a72582b87d89615e74167dec379dce1c6fd63a3c6b", "d0d6cf9b209b4c72600a014291700cee21e8a9d81a216e2205b884caff6447b2", "ac2840d34bc92c49aaef5f31a9b4fcefea84c551148a6e40c83eb32f3a3670b0"], "compiled_tokens_by_order": [407, 407, 407, 407, 407, 407], "choice": "einstein", "probs": {"darwin": 0.023003901647318886, "einstein": 0.8109154854444759, "hypatia": 0.007015793615556212, "ibn_sina": 0.014852435200683597, "sagan": 0.05874252074090991, "socrates": 0.08546986335105554}, "confidence": 0.5921417340160366, "seconds": 0.18151662616643685, "seconds_all_orders": 1.0890997569986212, "seconds_by_order": [0.17783822599812993, 0.18460702399897855, 0.1862682409991976, 0.17902384699846152, 0.1781632640013413, 0.18319915500251227], "argmax_stable": true, "spread": {"darwin": 0.09093025838660268, "einstein": 0.2919511697473701, "hypatia": 0.04925042475907489, "ibn_sina": 0.01823622289614774, "sagan": 0.06427371367158013, "socrates": 0.165467196024508}, "spread_2dp": {"darwin": 0.09, "einstein": 0.29000000000000004, "hypatia": 0.05, "ibn_sina": 0.019999999999999997, "sagan": 0.07, "socrates": 0.17}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.023003901647318886, "einstein": 0.8109154854444759, "hypatia": 0.007015793615556212, "ibn_sina": 0.014852435200683597, "sagan": 0.05874252074090991, "socrates": 0.08546986335105554}, {"ibn_sina": 0.01923309995725375, "hypatia": 0.004568261655591732, "sagan": 0.006646781488615499, "socrates": 0.0433424598196202, "darwin": 0.05565282003015951, "einstein": 0.8705565770487593}, {"hypatia": 0.014508293763541626, "darwin": 0.02710506117297577, "sagan": 0.02710506117297577, "ibn_sina": 0.021109442866711858, "einstein": 0.7441335114435764, "socrates": 0.16603862958021862}, {"ibn_sina": 0.028807066971330026, "einstein": 0.5786054073013892, "darwin": 0.11393416003392157, "sagan": 0.03698900617141371, "hypatia": 0.05381868641466662, "socrates": 0.1878456731072789}, {"einstein": 0.8336710749815677, "sagan": 0.004656848491896812, "socrates": 0.036628950685219853, "ibn_sina": 0.01730229116569793, "darwin": 0.09956781104315561, "hypatia": 0.008173023632462043}, {"socrates": 0.022378477082770895, "ibn_sina": 0.010570844075182287, "hypatia": 0.00469078362364866, "einstein": 0.8397461572644532, "sagan": 0.06893056216347694, "darwin": 0.053683175790467996}], "prompt_tokens": 407, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.023003901647318886, "einstein": 0.8109154854444759, "hypatia": 0.007015793615556212, "ibn_sina": 0.014852435200683597, "sagan": 0.05874252074090991, "socrates": 0.08546986335105554}, "confidence": 0.5921417340160366, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.80956600108766, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "7f48d90264fa637fb31af7ae8698c998de26360d2468e492bee50203c0985399", "compiled_ids_sha256": "d0e8df03971a580df0512cfa9b64c5331720d7183da1c0497ce954a5948dbfa7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 19.25, 14.5, 15.25, 16.625, 17.0], "runtime_probabilities": [0.023003900423645973, 0.8109154105186462, 0.007015793118625879, 0.014852434396743774, 0.05874251946806908, 0.08546985685825348], "decide_seconds": 0.17314231199998176, "probabilities_f64_minus_runtime_max_abs": 7.492582965173966e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01923309995725375, "hypatia": 0.004568261655591732, "sagan": 0.006646781488615499, "socrates": 0.0433424598196202, "darwin": 0.05565282003015951, "einstein": 0.8705565770487593}, "confidence": 0.6922534529909088, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 184.5802839998214, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "8d80177bfd46c0d4a66807221004faa3440fa2c323586ff0418a3061b0726218", "compiled_ids_sha256": "c5e49280a42ef47696d79f747609618696042a3da2bb4d64a56325287de5f62e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 13.875, 14.25, 16.125, 16.375, 19.125], "runtime_probabilities": [0.0192331001162529, 0.004568262025713921, 0.0066467816941440105, 0.043342459946870804, 0.05565282329916954, 0.870556652545929], "decide_seconds": 0.17982040099741425, "probabilities_f64_minus_runtime_max_abs": 7.549716962707009e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.014508293763541626, "darwin": 0.02710506117297577, "sagan": 0.02710506117297577, "ibn_sina": 0.021109442866711858, "einstein": 0.7441335114435764, "socrates": 0.16603862958021862}, "confidence": 0.521982389019936, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 186.23408100029337, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "88d56b8abe9c5079891b409265cd812ec9c063b6e44d366648d272241bca7f80", "compiled_ids_sha256": "28ba71f0edb6fa32de922d92e82615c9c6b6d9df91df0d03e0e636ffcbc1937e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 15.8125, 15.8125, 15.5625, 19.125, 17.625], "runtime_probabilities": [0.014508293941617012, 0.0271050613373518, 0.0271050613373518, 0.0211094431579113, 0.744133472442627, 0.1660386323928833], "decide_seconds": 0.18138387900035013, "probabilities_f64_minus_runtime_max_abs": 3.900094946995836e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.028807066971330026, "einstein": 0.5786054073013892, "darwin": 0.11393416003392157, "sagan": 0.03698900617141371, "hypatia": 0.05381868641466662, "socrates": 0.1878456731072789}, "confidence": 0.29702378971886234, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 178.99780700099654, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "c52b8e78a74d935c029bc1a72582b87d89615e74167dec379dce1c6fd63a3c6b", "compiled_ids_sha256": "9b930df006e9d696ebe40b0aa154237df8c7117c0f6e84589c6dc8f79a2572c0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 19.0, 17.375, 16.25, 16.625, 17.875], "runtime_probabilities": [0.02880706638097763, 0.5786054134368896, 0.11393415182828903, 0.0369890071451664, 0.053818684071302414, 0.18784567713737488], "decide_seconds": 0.17424972300068475, "probabilities_f64_minus_runtime_max_abs": 8.205632534563634e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.8336710749815677, "sagan": 0.004656848491896812, "socrates": 0.036628950685219853, "ibn_sina": 0.01730229116569793, "darwin": 0.09956781104315561, "hypatia": 0.008173023632462043}, "confidence": 0.6445017072100385, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 178.13757399926544, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "d0d6cf9b209b4c72600a014291700cee21e8a9d81a216e2205b884caff6447b2", "compiled_ids_sha256": "b300f63c23e343fe72534d3c1d6e785100a3ff1628049c4a29bbc4d1471ce10d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 14.0625, 16.125, 15.375, 17.125, 14.625], "runtime_probabilities": [0.8336710929870605, 0.004656848963350058, 0.036628954112529755, 0.017302291467785835, 0.09956781566143036, 0.008173024281859398], "decide_seconds": 0.17352483899958315, "probabilities_f64_minus_runtime_max_abs": 1.800549287445108e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.022378477082770895, "ibn_sina": 0.010570844075182287, "hypatia": 0.00469078362364866, "einstein": 0.8397461572644532, "sagan": 0.06893056216347694, "darwin": 0.053683175790467996}, "confidence": 0.6392854963061119, "choice": "einstein"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 183.1726750024245, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "ac2840d34bc92c49aaef5f31a9b4fcefea84c551148a6e40c83eb32f3a3670b0", "compiled_ids_sha256": "b1c458ac84825aff6d52fdb309b3952cec0fe03f5c1727df32fcef32192f0065", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 14.875, 14.0625, 19.25, 16.75, 16.5], "runtime_probabilities": [0.022378476336598396, 0.010570843704044819, 0.004690783564001322, 0.8397461175918579, 0.06893055886030197, 0.05368317291140556], "decide_seconds": 0.17841505199976382, "probabilities_f64_minus_runtime_max_abs": 3.967259532000611e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:17Z"} {"id": "c-einstein-01-main-01", "uid": "c023", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "prompt_sha256_by_order": ["18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c"], "request_sha256": "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "compiled_prompt_sha256_by_order": ["398a30a8df523f7677bfca21e4476fae04e1c93179ccaa5354f79989bf208257", "fb796abcfc1335e000686da4402638395dd61ebcb112ceb23cb18348e2aba549", "1b8f31e83678fceb562733937f7c4aa1690be9e6eca1423bddb061547e47f722", "3247b7295dcf1bab300f2040053fc0c5881ed636abc78e6ba4ab4ebd2b08aad7", "b3b0cf383d3ca10ad0c5f751fe4ae2e1c49ca29042b183a4ccaeaa516d4ef2cd", "293bb653e9994fc9f04a10c056b0008f872b037462957434f1ff640f7bd66ba3"], "compiled_tokens_by_order": [299, 299, 299, 299, 299, 299], "choice": "sagan", "probs": {"darwin": 0.013994907110577838, "einstein": 0.04884702547373801, "hypatia": 0.04884702547373801, "ibn_sina": 0.007037081225968598, "sagan": 0.764096038123246, "socrates": 0.11717792259273159}, "confidence": 0.527617964599584, "seconds": 0.1662653763335887, "seconds_all_orders": 0.9975922580015322, "seconds_by_order": [0.16613026599952718, 0.1655696900015755, 0.16580679799881182, 0.16893901400180766, 0.16583756799809635, 0.16530892200171365], "argmax_stable": false, "spread": {"darwin": 0.06923608865867993, "einstein": 0.6102302259642338, "hypatia": 0.03153599363436127, "ibn_sina": 0.018601911301714805, "sagan": 0.757013956550301, "socrates": 0.22655932241525123}, "spread_2dp": {"darwin": 0.07, "einstein": 0.61, "hypatia": 0.030000000000000002, "ibn_sina": 0.02, "sagan": 0.76, "socrates": 0.22999999999999998}, "choices_by_order": ["sagan", "einstein", "einstein", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.013994907110577838, "einstein": 0.04884702547373801, "hypatia": 0.04884702547373801, "ibn_sina": 0.007037081225968598, "sagan": 0.764096038123246, "socrates": 0.11717792259273159}, {"ibn_sina": 0.02240131861238988, "hypatia": 0.021044091328725693, "sagan": 0.13722471310632764, "socrates": 0.12110038427439387, "darwin": 0.08323099576925777, "einstein": 0.6149984969089052}, {"hypatia": 0.017311031839376743, "darwin": 0.01842749747116718, "sagan": 0.06042143806696632, "ibn_sina": 0.015276931978793285, "einstein": 0.6495916773796507, "socrates": 0.23897142326404575}, {"ibn_sina": 0.011277244641896268, "einstein": 0.03936145141541692, "darwin": 0.03065472918515376, "sagan": 0.6976984203745716, "hypatia": 0.044602367781844725, "socrates": 0.17640578660111678}, {"einstein": 0.13393372441654738, "sagan": 0.7707353689348776, "socrates": 0.023274191452868603, "ibn_sina": 0.003799407310675074, "darwin": 0.02988465337833988, "hypatia": 0.03837265450669159}, {"socrates": 0.01241210084879453, "ibn_sina": 0.006643718831247531, "hypatia": 0.035915604674722726, "einstein": 0.11062785065816015, "sagan": 0.8174353946172673, "darwin": 0.016965330369807845}], "prompt_tokens": 299, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.013994907110577838, "einstein": 0.04884702547373801, "hypatia": 0.04884702547373801, "ibn_sina": 0.007037081225968598, "sagan": 0.764096038123246, "socrates": 0.11717792259273159}, "confidence": 0.527617964599584, "choice": "sagan"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 166.10095600117347, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "398a30a8df523f7677bfca21e4476fae04e1c93179ccaa5354f79989bf208257", "compiled_ids_sha256": "4e8ff629b9c571ecae098a26e5b2d5c09709bf9dbea7851caa35fd43fed66dce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 16.5, 16.5, 14.5625, 19.25, 17.375], "runtime_probabilities": [0.013994907028973103, 0.0488470196723938, 0.0488470196723938, 0.007037080358713865, 0.7640960216522217, 0.11717791855335236], "decide_seconds": 0.16228114600016852, "probabilities_f64_minus_runtime_max_abs": 1.64710243222288e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02240131861238988, "hypatia": 0.021044091328725693, "sagan": 0.13722471310632764, "socrates": 0.12110038427439387, "darwin": 0.08323099576925777, "einstein": 0.6149984969089052}, "confidence": 0.3300156423740037, "choice": "einstein"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 165.54454000288388, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "fb796abcfc1335e000686da4402638395dd61ebcb112ceb23cb18348e2aba549", "compiled_ids_sha256": "89d17dbea35c0b40e89d1602d364d14a2483e687edf0a8b96361ac1944d79cb8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.5, 17.375, 17.25, 16.875, 18.875], "runtime_probabilities": [0.022401319816708565, 0.021044092252850533, 0.1372247189283371, 0.12110038101673126, 0.08323099464178085, 0.6149985194206238], "decide_seconds": 0.16173858899855986, "probabilities_f64_minus_runtime_max_abs": 2.251171860034873e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.017311031839376743, "darwin": 0.01842749747116718, "sagan": 0.06042143806696632, "ibn_sina": 0.015276931978793285, "einstein": 0.6495916773796507, "socrates": 0.23897142326404575}, "confidence": 0.44212791780847105, "choice": "einstein"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 165.78048900191789, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "1b8f31e83678fceb562733937f7c4aa1690be9e6eca1423bddb061547e47f722", "compiled_ids_sha256": "7f77fd5c86cfca37884acf2197ce451c9662901c9e6acd77875110527e99ebec", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 15.5625, 16.75, 15.375, 19.125, 18.125], "runtime_probabilities": [0.017311030998826027, 0.018427496775984764, 0.060421429574489594, 0.01527693122625351, 0.6495916247367859, 0.23897142708301544], "decide_seconds": 0.16211306599871023, "probabilities_f64_minus_runtime_max_abs": 5.2642864845076076e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.011277244641896268, "einstein": 0.03936145141541692, "darwin": 0.03065472918515376, "sagan": 0.6976984203745716, "hypatia": 0.044602367781844725, "socrates": 0.17640578660111678}, "confidence": 0.45268232004593023, "choice": "sagan"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 168.91454399956274, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "3247b7295dcf1bab300f2040053fc0c5881ed636abc78e6ba4ab4ebd2b08aad7", "compiled_ids_sha256": "267064f650ecc4e24294fcecd4900507c9af3237fcd9d02a31158c23fb67dd9f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 16.25, 16.0, 19.125, 16.375, 17.75], "runtime_probabilities": [0.011277244426310062, 0.03936145454645157, 0.030654728412628174, 0.6976984143257141, 0.04460236430168152, 0.17640578746795654], "decide_seconds": 0.1652575820007769, "probabilities_f64_minus_runtime_max_abs": 6.048857525975393e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.13393372441654738, "sagan": 0.7707353689348776, "socrates": 0.023274191452868603, "ibn_sina": 0.003799407310675074, "darwin": 0.02988465337833988, "hypatia": 0.03837265450669159}, "confidence": 0.5486663610101958, "choice": "sagan"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 165.81309800312738, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "b3b0cf383d3ca10ad0c5f751fe4ae2e1c49ca29042b183a4ccaeaa516d4ef2cd", "compiled_ids_sha256": "00b244220fca4faeb874f282a9c1621ba27fca6bedca2049e1f9f71a780b506a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.75, 19.5, 16.0, 14.1875, 16.25, 16.5], "runtime_probabilities": [0.13393372297286987, 0.7707354426383972, 0.023274192586541176, 0.003799407510086894, 0.02988465502858162, 0.03837265446782112], "decide_seconds": 0.16212365699902875, "probabilities_f64_minus_runtime_max_abs": 7.370351962077137e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.01241210084879453, "ibn_sina": 0.006643718831247531, "hypatia": 0.035915604674722726, "einstein": 0.11062785065816015, "sagan": 0.8174353946172673, "darwin": 0.016965330369807845}, "confidence": 0.6178256013865115, "choice": "sagan"}}, "usage": {"input_tokens": 299, "output_tokens": 1}, "latency_ms": 165.28398200171068, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 299, "compiled_prompt_sha256": "293bb653e9994fc9f04a10c056b0008f872b037462957434f1ff640f7bd66ba3", "compiled_ids_sha256": "f9154e3a4b7bc6bab19845d1da15fc6951a0f6caf7ce89a37aab4e51345472e6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.4375, 16.125, 17.25, 19.25, 15.375], "runtime_probabilities": [0.012412101030349731, 0.006643719505518675, 0.03591560572385788, 0.11062785238027573, 0.8174353837966919, 0.01696532964706421], "decide_seconds": 0.16160321999996086, "probabilities_f64_minus_runtime_max_abs": 1.0820575435666058e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:18Z"} {"id": "c-einstein-01-main-01", "uid": "c024", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "prompt_sha256_by_order": ["0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2"], "request_sha256": "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "compiled_prompt_sha256_by_order": ["31ad7358200e1676880e3e7daab4375eb6f0352b2bc77278e208e561783c83f2", "e0c2f8e76dfd6c1703219e9e816b92c0098057c8e124d4b6a0f6d23a741ad3fc", "d1e86d115749e78a173b8a21a6b94e16f0e4d7bb92c0f51751a4ac82a7a313fa", "47a9372a4aa4f3f9bf55c5958e2d6a15407f66a6330504bd4ed931d10e3aa32a", "9703b21f7fe24b0adf4d457b162e354446267f4af9f4a26da7f51d3dbbde02a8", "b4f413074fb81b8da1b697c2898480d0882dfbb137eee1f36bcb897aabcb1905"], "compiled_tokens_by_order": [346, 346, 346, 346, 346, 346], "choice": "einstein", "probs": {"darwin": 0.015603445618679821, "einstein": 0.6634753906354051, "hypatia": 0.014658080639087521, "ibn_sina": 0.06171282456379763, "sagan": 0.190088882015732, "socrates": 0.05446137652729795}, "confidence": 0.41677837357437686, "seconds": 0.17459631233396067, "seconds_all_orders": 1.047577874003764, "seconds_by_order": [0.17181548200096586, 0.1791377059998922, 0.17858266999974148, 0.17070320100174285, 0.17676196400134359, 0.17057685100007802], "argmax_stable": true, "spread": {"darwin": 0.019194515475635953, "einstein": 0.22392565679854481, "hypatia": 0.05988958988170384, "ibn_sina": 0.0558393667923845, "sagan": 0.17232196269099787, "socrates": 0.15593991015163855}, "spread_2dp": {"darwin": 0.019999999999999997, "einstein": 0.22999999999999998, "hypatia": 0.06, "ibn_sina": 0.05, "sagan": 0.17, "socrates": 0.16}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.015603445618679821, "einstein": 0.6634753906354051, "hypatia": 0.014658080639087521, "ibn_sina": 0.06171282456379763, "sagan": 0.190088882015732, "socrates": 0.05446137652729795}, {"ibn_sina": 0.028921864771466098, "hypatia": 0.012056426961502059, "sagan": 0.021159688285085892, "socrates": 0.16643384040895468, "darwin": 0.025523456077789358, "einstein": 0.745904723495202}, {"hypatia": 0.010434658991767997, "darwin": 0.006328940602153406, "sagan": 0.05299157087027691, "ibn_sina": 0.06004731655723537, "einstein": 0.8289276360886108, "socrates": 0.04126987688995551}, {"ibn_sina": 0.05004893984802035, "einstein": 0.6909043030660066, "darwin": 0.01841197602251286, "sagan": 0.0222090824299688, "hypatia": 0.06426411084313405, "socrates": 0.15416158779035732}, {"einstein": 0.8041921742422288, "sagan": 0.01776691932473414, "socrates": 0.05825548571095346, "ibn_sina": 0.0847612315638506, "darwin": 0.02281317598919924, "hypatia": 0.012211013169033915}, {"socrates": 0.010493930257316124, "ibn_sina": 0.030365194030991256, "hypatia": 0.004374520961430209, "einstein": 0.8874010474339499, "sagan": 0.05006374128783186, "darwin": 0.01730156602848076}], "prompt_tokens": 346, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.015603445618679821, "einstein": 0.6634753906354051, "hypatia": 0.014658080639087521, "ibn_sina": 0.06171282456379763, "sagan": 0.190088882015732, "socrates": 0.05446137652729795}, "confidence": 0.41677837357437686, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 171.78573199998937, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "31ad7358200e1676880e3e7daab4375eb6f0352b2bc77278e208e561783c83f2", "compiled_ids_sha256": "c9faf3bf2ed711f2c81c34c713922081a5f5829813476075c406186f3052df81", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 19.125, 15.3125, 16.75, 17.875, 16.625], "runtime_probabilities": [0.015603446401655674, 0.6634753942489624, 0.014658080413937569, 0.06171282380819321, 0.1900888830423355, 0.05446137487888336], "decide_seconds": 0.1676316639968718, "probabilities_f64_minus_runtime_max_abs": 3.613557342241336e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.028921864771466098, "hypatia": 0.012056426961502059, "sagan": 0.021159688285085892, "socrates": 0.16643384040895468, "darwin": 0.025523456077789358, "einstein": 0.745904723495202}, "confidence": 0.5266886799768887, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 179.1083459975198, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "e0c2f8e76dfd6c1703219e9e816b92c0098057c8e124d4b6a0f6d23a741ad3fc", "compiled_ids_sha256": "8f7149783acf5f26b2416319ec825186cc44b96cf2da6456b8aacab1480206b9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 15.25, 15.8125, 17.875, 16.0, 19.375], "runtime_probabilities": [0.028921866789460182, 0.012056428007781506, 0.021159691736102104, 0.16643385589122772, 0.025523457676172256, 0.745904803276062], "decide_seconds": 0.17472118899968336, "probabilities_f64_minus_runtime_max_abs": 7.978086002946583e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.010434658991767997, "darwin": 0.006328940602153406, "sagan": 0.05299157087027691, "ibn_sina": 0.06004731655723537, "einstein": 0.8289276360886108, "socrates": 0.04126987688995551}, "confidence": 0.6141846574920669, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 178.55495100229746, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "d1e86d115749e78a173b8a21a6b94e16f0e4d7bb92c0f51751a4ac82a7a313fa", "compiled_ids_sha256": "ba200ceb0620aa31241c2cd17a3d157bc37d9433ee04195900e2e5b7a11ef412", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 14.625, 16.75, 16.875, 19.5, 16.5], "runtime_probabilities": [0.010434658266603947, 0.00632893992587924, 0.05299156531691551, 0.060047317296266556, 0.8289275765419006, 0.04126987233757973], "decide_seconds": 0.17438734199822647, "probabilities_f64_minus_runtime_max_abs": 5.954671011831891e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.05004893984802035, "einstein": 0.6909043030660066, "darwin": 0.01841197602251286, "sagan": 0.0222090824299688, "hypatia": 0.06426411084313405, "socrates": 0.15416158779035732}, "confidence": 0.4262124917249226, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 170.6757510000898, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "47a9372a4aa4f3f9bf55c5958e2d6a15407f66a6330504bd4ed931d10e3aa32a", "compiled_ids_sha256": "f7003fc8d27823fc8645d3ec99b5a66135b4cf57b491b6aee4da2a8691d0a385", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 19.25, 15.625, 15.8125, 16.875, 17.75], "runtime_probabilities": [0.050048939883708954, 0.6909042596817017, 0.018411975353956223, 0.022209081798791885, 0.06426410377025604, 0.15416158735752106], "decide_seconds": 0.16656493200207478, "probabilities_f64_minus_runtime_max_abs": 4.33843049263416e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.8041921742422288, "sagan": 0.01776691932473414, "socrates": 0.05825548571095346, "ibn_sina": 0.0847612315638506, "darwin": 0.02281317598919924, "hypatia": 0.012211013169033915}, "confidence": 0.5748913920652077, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 176.73537400332862, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "9703b21f7fe24b0adf4d457b162e354446267f4af9f4a26da7f51d3dbbde02a8", "compiled_ids_sha256": "7c92edd32eeb651990329eadde74066cb997508e2801de62eab97e40a22c8ff1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 15.5625, 16.75, 17.125, 15.8125, 15.1875], "runtime_probabilities": [0.8041922450065613, 0.0177669208496809, 0.05825549364089966, 0.08476123213768005, 0.022813178598880768, 0.012211013585329056], "decide_seconds": 0.1726494750000711, "probabilities_f64_minus_runtime_max_abs": 7.076433250485792e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.010493930257316124, "ibn_sina": 0.030365194030991256, "hypatia": 0.004374520961430209, "einstein": 0.8874010474339499, "sagan": 0.05006374128783186, "darwin": 0.01730156602848076}, "confidence": 0.7188208413188912, "choice": "einstein"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 170.55013199933455, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "b4f413074fb81b8da1b697c2898480d0882dfbb137eee1f36bcb897aabcb1905", "compiled_ids_sha256": "3898042e21451135a44b30cd6af9147537d0ac85cb9072a829ace1430945ff80", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 16.25, 14.3125, 19.625, 16.75, 15.6875], "runtime_probabilities": [0.0104939304292202, 0.030365195125341415, 0.004374521318823099, 0.8874011039733887, 0.05006374791264534, 0.017301568761467934], "decide_seconds": 0.16644266399816843, "probabilities_f64_minus_runtime_max_abs": 5.6539438797464925e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:19Z"} {"id": "c-einstein-01-main-02", "uid": "c025", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "prompt_sha256_by_order": ["af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11"], "request_sha256": "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "compiled_prompt_sha256_by_order": ["8bc0000acb4ce0e5f74190578a430fa97ca4b3961d56af5f977ae0c41db74aee", "0494bc9fefd8d4c25ac86fdcaef4fbc76617fddf7ec9a239458b2053f77e14e5", "76ed487276a05ea6bf05d42b96ee8cd806fd097a7505da0d32b0d4a3b181c789", "00f01aac12f5e659cbc7bc3aba83f2121260446d262a1b7f39663aff3aff6f3b", "40382364013268e8833268571f4b4cceee955f64fe25674b6d23848afe4523f1", "7d295551f16192316aaa0e865f85be295172fe2d1500426daebd57fa6cb99971"], "compiled_tokens_by_order": [384, 384, 384, 384, 384, 384], "choice": "sagan", "probs": {"darwin": 0.006328034410271629, "einstein": 0.02502786159790576, "hypatia": 0.009801051425428076, "ibn_sina": 0.0024780938808667695, "sagan": 0.9391635777382228, "socrates": 0.017201380947304917}, "confidence": 0.8251058304350644, "seconds": 0.17364580999916748, "seconds_all_orders": 1.041874859995005, "seconds_by_order": [0.17397556599826203, 0.17132314599803067, 0.17427184299958753, 0.17289693400016404, 0.17870884999865666, 0.17069852100030403], "argmax_stable": true, "spread": {"darwin": 0.004212310382502909, "einstein": 0.06710649377918042, "hypatia": 0.006312812217914439, "ibn_sina": 0.0031155578234610052, "sagan": 0.13873876194324286, "socrates": 0.07213360538987743}, "spread_2dp": {"darwin": 0.01, "einstein": 0.07, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.14, "socrates": 0.07}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.006328034410271629, "einstein": 0.02502786159790576, "hypatia": 0.009801051425428076, "ibn_sina": 0.0024780938808667695, "sagan": 0.9391635777382228, "socrates": 0.017201380947304917}, {"ibn_sina": 0.005593651704327775, "hypatia": 0.0038444568456753284, "sagan": 0.8301715203587968, "socrates": 0.07721797992341341, "darwin": 0.005954411244373356, "einstein": 0.07721797992341341}, {"hypatia": 0.003488239207513637, "darwin": 0.00211572402776872, "sagan": 0.9671920540304726, "ibn_sina": 0.002552050321721425, "einstein": 0.017714740397675212, "socrates": 0.006937192014848409}, {"ibn_sina": 0.0037198078721532163, "einstein": 0.010111486144232995, "darwin": 0.004776327853039874, "sagan": 0.9689102823020397, "hypatia": 0.007397721294998276, "socrates": 0.005084374533535983}, {"einstein": 0.03404717849561534, "sagan": 0.8780883066232655, "socrates": 0.07207787744083238, "ibn_sina": 0.002794762599960222, "darwin": 0.004904961116158659, "hypatia": 0.008086913724167824}, {"socrates": 0.0073697931940440815, "ibn_sina": 0.002546932376266359, "hypatia": 0.008351069757253476, "einstein": 0.01376858634188436, "sagan": 0.9652524229287777, "darwin": 0.0027111954017740355}], "prompt_tokens": 384, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.006328034410271629, "einstein": 0.02502786159790576, "hypatia": 0.009801051425428076, "ibn_sina": 0.0024780938808667695, "sagan": 0.9391635777382228, "socrates": 0.017201380947304917}, "confidence": 0.8251058304350644, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 173.94712600071216, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "8bc0000acb4ce0e5f74190578a430fa97ca4b3961d56af5f977ae0c41db74aee", "compiled_ids_sha256": "96c01d22640684d10ebbdf561b48b5676cfc3b4bf58f2c5ccd9550d3ab256a3e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 15.75, 14.8125, 13.4375, 19.375, 15.375], "runtime_probabilities": [0.0063280342146754265, 0.025027859956026077, 0.009801051579415798, 0.0024780938401818275, 0.9391635656356812, 0.0172013808041811], "decide_seconds": 0.1693959399999585, "probabilities_f64_minus_runtime_max_abs": 1.2102541635528041e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.005593651704327775, "hypatia": 0.0038444568456753284, "sagan": 0.8301715203587968, "socrates": 0.07721797992341341, "darwin": 0.005954411244373356, "einstein": 0.07721797992341341}, "confidence": 0.6478654174488823, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 171.29696600022726, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "0494bc9fefd8d4c25ac86fdcaef4fbc76617fddf7ec9a239458b2053f77e14e5", "compiled_ids_sha256": "4a8956bac8ac84f21e010cb16becbc7f4be64a5c282e86a2d016ebd22a3d9d59", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 14.0, 19.375, 17.0, 14.4375, 17.0], "runtime_probabilities": [0.005593651905655861, 0.0038444572128355503, 0.8301715850830078, 0.07721798121929169, 0.005954411346465349, 0.07721798121929169], "decide_seconds": 0.16669364100016537, "probabilities_f64_minus_runtime_max_abs": 6.47242109907964e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.003488239207513637, "darwin": 0.00211572402776872, "sagan": 0.9671920540304726, "ibn_sina": 0.002552050321721425, "einstein": 0.017714740397675212, "socrates": 0.006937192014848409}, "confidence": 0.8960784081175088, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 174.24620300153038, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "76ed487276a05ea6bf05d42b96ee8cd806fd097a7505da0d32b0d4a3b181c789", "compiled_ids_sha256": "b3bf1be97c87930b57ae8957c6dc8281df31600a11862e149815ef7e9466c54f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 13.5, 19.625, 13.6875, 15.625, 14.6875], "runtime_probabilities": [0.0034882391337305307, 0.0021157239098101854, 0.9671919941902161, 0.0025520501658320427, 0.017714738845825195, 0.006937190890312195], "decide_seconds": 0.1697010190000583, "probabilities_f64_minus_runtime_max_abs": 5.984025652772118e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0037198078721532163, "einstein": 0.010111486144232995, "darwin": 0.004776327853039874, "sagan": 0.9689102823020397, "hypatia": 0.007397721294998276, "socrates": 0.005084374533535983}, "confidence": 0.8958903532195502, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 172.87159400075325, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "00f01aac12f5e659cbc7bc3aba83f2121260446d262a1b7f39663aff3aff6f3b", "compiled_ids_sha256": "43d9a73f26bee34e62258a74f4dfa19b595076038387a93a74238d4108528dae", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.9375, 14.9375, 14.1875, 19.5, 14.625, 14.25], "runtime_probabilities": [0.0037198083009570837, 0.010111487470567226, 0.004776328336447477, 0.9689103960990906, 0.007397721521556377, 0.0050843749195337296], "decide_seconds": 0.16840815799878328, "probabilities_f64_minus_runtime_max_abs": 1.137970508935382e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03404717849561534, "sagan": 0.8780883066232655, "socrates": 0.07207787744083238, "ibn_sina": 0.002794762599960222, "darwin": 0.004904961116158659, "hypatia": 0.008086913724167824}, "confidence": 0.7207895788963616, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 178.68358999839984, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "40382364013268e8833268571f4b4cceee955f64fe25674b6d23848afe4523f1", "compiled_ids_sha256": "3949bd5cec5d7bf94d9f984527eb8d1d4729b1390d3aab6fe96ef9bd24b0e01d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 19.625, 17.125, 13.875, 14.4375, 14.9375], "runtime_probabilities": [0.03404717892408371, 0.8780883550643921, 0.07207787781953812, 0.0027947628404945135, 0.004904961679130793, 0.00808691419661045], "decide_seconds": 0.17420856399985496, "probabilities_f64_minus_runtime_max_abs": 4.844112655444377e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0073697931940440815, "ibn_sina": 0.002546932376266359, "hypatia": 0.008351069757253476, "einstein": 0.01376858634188436, "sagan": 0.9652524229287777, "darwin": 0.0027111954017740355}, "confidence": 0.8880831181759887, "choice": "sagan"}}, "usage": {"input_tokens": 384, "output_tokens": 1}, "latency_ms": 170.67305100135854, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 384, "compiled_prompt_sha256": "7d295551f16192316aaa0e865f85be295172fe2d1500426daebd57fa6cb99971", "compiled_ids_sha256": "2cce23a5300473a30b8b4c1d5991f9b53b5b8a138d40cadb5e63259074bbd485", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 13.5625, 14.75, 15.25, 19.5, 13.625], "runtime_probabilities": [0.007369792554527521, 0.002546932315453887, 0.008351069875061512, 0.01376858539879322, 0.9652523994445801, 0.0027111954987049103], "decide_seconds": 0.1661563150009897, "probabilities_f64_minus_runtime_max_abs": 2.3484197586576272e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:20Z"} {"id": "c-einstein-01-main-03", "uid": "c026", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "prompt_sha256_by_order": ["fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd"], "request_sha256": "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "compiled_prompt_sha256_by_order": ["babae6f4edd94f8ec21e9cf611f98b3fccee444a918dac8ccf6454902374a632", "1184b761fd135b39e09455ad71e100141129446e3b27aa3aaec8a1e9e2f23378", "48e09c0736f5871ec68014b23b8b65dcbfdf6135a77b7fe34ef3001271e60c91", "129ed980177a61906be57f39da63f7afb936fdd51f73e537dea1d40e65b51681", "f6768315e2c58c3e890e1ba6bc6ab8f3937fcbf10b7475df64e2b254f1fa389b", "d76a2f91dac671039feb16b900913661869267184bfad121f33add519ad4a47f"], "compiled_tokens_by_order": [339, 339, 339, 339, 339, 339], "choice": "sagan", "probs": {"darwin": 0.007359403209833365, "einstein": 0.039784557871879137, "hypatia": 0.027343500086781505, "ibn_sina": 0.005384254408034171, "sagan": 0.9054923635063121, "socrates": 0.014635920917159801}, "confidence": 0.7529299875472125, "seconds": 0.1732363778337458, "seconds_all_orders": 1.0394182670024747, "seconds_by_order": [0.17274675500084413, 0.17478004900112865, 0.1776777069972013, 0.17248827700313996, 0.17092997900181217, 0.17079549999834853], "argmax_stable": true, "spread": {"darwin": 0.01510312084341493, "einstein": 0.1704746738532821, "hypatia": 0.0608505184320156, "ibn_sina": 0.01843451426048945, "sagan": 0.23935849042384183, "socrates": 0.1130026352010487}, "spread_2dp": {"darwin": 0.01, "einstein": 0.17, "hypatia": 0.06, "ibn_sina": 0.01, "sagan": 0.24, "socrates": 0.11}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.007359403209833365, "einstein": 0.039784557871879137, "hypatia": 0.027343500086781505, "ibn_sina": 0.005384254408034171, "sagan": 0.9054923635063121, "socrates": 0.014635920917159801}, {"ibn_sina": 0.02055182464123736, "hypatia": 0.033884230437707066, "sagan": 0.6805829615703647, "socrates": 0.05586565146395236, "darwin": 0.014125048735514557, "einstein": 0.1949902831512239}, {"hypatia": 0.00975526733098622, "darwin": 0.005916868729936174, "sagan": 0.8781411802050777, "ibn_sina": 0.012526011199569895, "einstein": 0.06361233373582666, "socrates": 0.03004833879860339}, {"ibn_sina": 0.02381876866852362, "einstein": 0.07336697257326735, "darwin": 0.021019989573351102, "sagan": 0.6960864548885217, "hypatia": 0.06474612604791741, "socrates": 0.12096168824841881}, {"einstein": 0.06905129720265557, "sagan": 0.8412170111500921, "socrates": 0.06905129720265557, "ibn_sina": 0.006837004287978472, "darwin": 0.009947782540716505, "hypatia": 0.0038956076159018063}, {"socrates": 0.007959053047370108, "ibn_sina": 0.00747683840032488, "hypatia": 0.027779824751953463, "einstein": 0.024515609297941807, "sagan": 0.9199414519942065, "darwin": 0.01232722250820315}], "prompt_tokens": 339, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.007359403209833365, "einstein": 0.039784557871879137, "hypatia": 0.027343500086781505, "ibn_sina": 0.005384254408034171, "sagan": 0.9054923635063121, "socrates": 0.014635920917159801}, "confidence": 0.7529299875472125, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 172.7169350015174, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "babae6f4edd94f8ec21e9cf611f98b3fccee444a918dac8ccf6454902374a632", "compiled_ids_sha256": "6866bdd0a49c4b3aa6a808c4deba4a1c9165cab5449d0a129281ece86c4bd505", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 16.5, 16.125, 14.5, 19.625, 15.5], "runtime_probabilities": [0.0073594036512076855, 0.03978455811738968, 0.027343500405550003, 0.005384254269301891, 0.90549236536026, 0.014635920524597168], "decide_seconds": 0.16851996699915617, "probabilities_f64_minus_runtime_max_abs": 1.8539478841006485e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02055182464123736, "hypatia": 0.033884230437707066, "sagan": 0.6805829615703647, "socrates": 0.05586565146395236, "darwin": 0.014125048735514557, "einstein": 0.1949902831512239}, "confidence": 0.44382815172759493, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 174.75146899960237, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "1184b761fd135b39e09455ad71e100141129446e3b27aa3aaec8a1e9e2f23378", "compiled_ids_sha256": "463355481f2dffe0a9c016bddd3af0e00894e36f2d49f66a397429da5cacae06", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 16.5, 19.5, 17.0, 15.625, 18.25], "runtime_probabilities": [0.02055182494223118, 0.03388422727584839, 0.6805829405784607, 0.05586564913392067, 0.01412504818290472, 0.19499027729034424], "decide_seconds": 0.17058653200001572, "probabilities_f64_minus_runtime_max_abs": 2.099190399285078e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.00975526733098622, "darwin": 0.005916868729936174, "sagan": 0.8781411802050777, "ibn_sina": 0.012526011199569895, "einstein": 0.06361233373582666, "socrates": 0.03004833879860339}, "confidence": 0.7069573101021203, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 177.65036699711345, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "48e09c0736f5871ec68014b23b8b65dcbfdf6135a77b7fe34ef3001271e60c91", "compiled_ids_sha256": "4e03e8345f39d057cb29a3a6063270e2508890706a054f5d717d30d2f3680912", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 14.875, 19.875, 15.625, 17.25, 16.5], "runtime_probabilities": [0.009755267761647701, 0.005916868802160025, 0.8781412243843079, 0.01252601109445095, 0.06361233443021774, 0.030048338696360588], "decide_seconds": 0.17341605000183336, "probabilities_f64_minus_runtime_max_abs": 4.4179230163088334e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02381876866852362, "einstein": 0.07336697257326735, "darwin": 0.021019989573351102, "sagan": 0.6960864548885217, "hypatia": 0.06474612604791741, "socrates": 0.12096168824841881}, "confidence": 0.41578631094356844, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 172.46105699814507, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "129ed980177a61906be57f39da63f7afb936fdd51f73e537dea1d40e65b51681", "compiled_ids_sha256": "05a1442043d0408e3840b5fd47a230f9ec1e5f7740b118681f2f7ee04f65e15d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 17.25, 16.0, 19.5, 17.125, 17.75], "runtime_probabilities": [0.023818770423531532, 0.07336697727441788, 0.021019991487264633, 0.6960865259170532, 0.06474612653255463, 0.12096168845891953], "decide_seconds": 0.16836167799920077, "probabilities_f64_minus_runtime_max_abs": 7.102853150087185e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.06905129720265557, "sagan": 0.8412170111500921, "socrates": 0.06905129720265557, "ibn_sina": 0.006837004287978472, "darwin": 0.009947782540716505, "hypatia": 0.0038956076159018063}, "confidence": 0.6561216119858461, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.90316899702884, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "f6768315e2c58c3e890e1ba6bc6ab8f3937fcbf10b7475df64e2b254f1fa389b", "compiled_ids_sha256": "146b49a177ff9a096200876aba7359d1ef49602baa41a718aaf8dcff1296c6eb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 19.625, 17.125, 14.8125, 15.1875, 14.25], "runtime_probabilities": [0.06905129551887512, 0.8412169814109802, 0.06905129551887512, 0.006837004330009222, 0.009947781451046467, 0.003895607776939869], "decide_seconds": 0.16679246100102318, "probabilities_f64_minus_runtime_max_abs": 2.973911183978828e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.007959053047370108, "ibn_sina": 0.00747683840032488, "hypatia": 0.027779824751953463, "einstein": 0.024515609297941807, "sagan": 0.9199414519942065, "darwin": 0.01232722250820315}, "confidence": 0.7787132362963127, "choice": "sagan"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.7681900006719, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "d76a2f91dac671039feb16b900913661869267184bfad121f33add519ad4a47f", "compiled_ids_sha256": "776c074e939bce142305e4890110d00b065f7e733c091c93a6373e0fe023172f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 14.8125, 16.125, 16.0, 19.625, 15.3125], "runtime_probabilities": [0.007959052920341492, 0.007476838771253824, 0.027779823169112206, 0.024515608325600624, 0.9199414253234863, 0.012327222153544426], "decide_seconds": 0.16663951100053964, "probabilities_f64_minus_runtime_max_abs": 2.667072018791572e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:22Z"} {"id": "c-einstein-01-main-03", "uid": "c027", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "prompt_sha256_by_order": ["7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af"], "request_sha256": "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "compiled_prompt_sha256_by_order": ["5b0777e33fab5b56af64797557c932888e6e56a8386aa298f23774efd807c843", "0d097107b6a8fe4f153aa29a3af80c9c138fd5d2816bebbbec8d60ed14143d0a", "08504ba6b8e244df702dd952429002075c657f9019924174db85f519dc7d4195", "975fb5a80fe63ff6e9b0cea5774e58f849b0b863b1fb38437460dafb81c77b98", "0479b973affcdcbaddd3c7ab9b3104ed54e2f719ac405e19ef024dec2504ae60", "c3c12676b722efd1aae12c93922ffc374ad0ba0d69b50bbaf70f9d7e70b298bf"], "compiled_tokens_by_order": [345, 345, 345, 345, 345, 345], "choice": "einstein", "probs": {"darwin": 0.006098332505440242, "einstein": 0.9634449724168069, "hypatia": 0.0057288532169905955, "ibn_sina": 0.004461635371493413, "sagan": 0.01139318505036625, "socrates": 0.008873021438902659}, "confidence": 0.8807853808346913, "seconds": 0.17240018583349107, "seconds_all_orders": 1.0344011150009464, "seconds_by_order": [0.17104566800117027, 0.1710978180017264, 0.17649917599919718, 0.17439302200000384, 0.1708088300001691, 0.17055660099867964], "argmax_stable": true, "spread": {"darwin": 0.006783082072529095, "einstein": 0.05148122405025135, "hypatia": 0.012695117488852586, "ibn_sina": 0.03089903654266765, "sagan": 0.021214669860076925, "socrates": 0.008496520039043247}, "spread_2dp": {"darwin": 0.01, "einstein": 0.04999999999999993, "hypatia": 0.01, "ibn_sina": 0.04, "sagan": 0.03, "socrates": 0.01}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.006098332505440242, "einstein": 0.9634449724168069, "hypatia": 0.0057288532169905955, "ibn_sina": 0.004461635371493413, "sagan": 0.01139318505036625, "socrates": 0.008873021438902659}, {"ibn_sina": 0.014278461599272332, "hypatia": 0.012600698135030921, "sagan": 0.01183726042861745, "socrates": 0.011120077074568276, "darwin": 0.009813433574808472, "einstein": 0.9403500691877025}, {"hypatia": 0.006026573578587985, "darwin": 0.0030303515022793766, "sagan": 0.025372822481577677, "ibn_sina": 0.004693500222241761, "einstein": 0.9521081393989028, "socrates": 0.008768612816410408}, {"ibn_sina": 0.013446098987659149, "einstein": 0.9426443139842268, "darwin": 0.003618968447179636, "sagan": 0.004646847468369621, "hypatia": 0.0183786382571111, "socrates": 0.017265132855453655}, {"einstein": 0.9558212267623126, "sagan": 0.004158152621500753, "socrates": 0.015449411226026707, "ibn_sina": 0.010618214699281045, "darwin": 0.00826947392262038, "hypatia": 0.005683520768258513}, {"socrates": 0.011479929384381488, "ibn_sina": 0.03536067191416106, "hypatia": 0.012220321218442406, "einstein": 0.9119637483665556, "sagan": 0.022830565714878237, "darwin": 0.006144763401581168}], "prompt_tokens": 345, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.006098332505440242, "einstein": 0.9634449724168069, "hypatia": 0.0057288532169905955, "ibn_sina": 0.004461635371493413, "sagan": 0.01139318505036625, "socrates": 0.008873021438902659}, "confidence": 0.8807853808346913, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 171.01798800285906, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "5b0777e33fab5b56af64797557c932888e6e56a8386aa298f23774efd807c843", "compiled_ids_sha256": "526e5ba6afca227b1070971298c7ab7274443e599e3f4513ee13c7dcebd97fa0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 19.375, 14.25, 14.0, 14.9375, 14.6875], "runtime_probabilities": [0.006098332814872265, 0.9634450078010559, 0.005728853400796652, 0.004461635835468769, 0.011393184773623943, 0.00887302216142416], "decide_seconds": 0.16681693099963013, "probabilities_f64_minus_runtime_max_abs": 3.5384248975312005e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.014278461599272332, "hypatia": 0.012600698135030921, "sagan": 0.01183726042861745, "socrates": 0.011120077074568276, "darwin": 0.009813433574808472, "einstein": 0.9403500691877025}, "confidence": 0.8205441186316293, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 171.0741879978741, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "0d097107b6a8fe4f153aa29a3af80c9c138fd5d2816bebbbec8d60ed14143d0a", "compiled_ids_sha256": "49cab2c0d062118149ff27befc49d89cea6a55ee4887505904f2f22ea8d67763", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.1875, 15.125, 15.0625, 14.9375, 19.5], "runtime_probabilities": [0.014278462156653404, 0.012600699439644814, 0.011837261728942394, 0.011120077222585678, 0.00981343537569046, 0.9403501749038696], "decide_seconds": 0.16699894900011714, "probabilities_f64_minus_runtime_max_abs": 1.0571616715360221e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.006026573578587985, "darwin": 0.0030303515022793766, "sagan": 0.025372822481577677, "ibn_sina": 0.004693500222241761, "einstein": 0.9521081393989028, "socrates": 0.008768612816410408}, "confidence": 0.8576682377857644, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 176.47463600224, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "08504ba6b8e244df702dd952429002075c657f9019924174db85f519dc7d4195", "compiled_ids_sha256": "ea20be630b9b05af076a10c4e9a222911b5099a3967043cea3f18ea5c9f47c59", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 13.875, 16.0, 14.3125, 19.625, 14.9375], "runtime_probabilities": [0.00602657301351428, 0.0030303513631224632, 0.0253728199750185, 0.004693499766290188, 0.9521080255508423, 0.008768611587584019], "decide_seconds": 0.17227852799987886, "probabilities_f64_minus_runtime_max_abs": 1.1384806053449381e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013446098987659149, "einstein": 0.9426443139842268, "darwin": 0.003618968447179636, "sagan": 0.004646847468369621, "hypatia": 0.0183786382571111, "socrates": 0.017265132855453655}, "confidence": 0.8311963811493481, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 174.36906199873192, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "975fb5a80fe63ff6e9b0cea5774e58f849b0b863b1fb38437460dafb81c77b98", "compiled_ids_sha256": "94209938a1eb12f2bd585124138c5f6f1aa76e3a2442bb11f86578a8721ae68e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 19.375, 13.8125, 14.0625, 15.4375, 15.375], "runtime_probabilities": [0.013446099124848843, 0.9426443576812744, 0.003618968650698662, 0.004646847490221262, 0.018378639593720436, 0.017265133559703827], "decide_seconds": 0.1702044539997587, "probabilities_f64_minus_runtime_max_abs": 4.369704764695115e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9558212267623126, "sagan": 0.004158152621500753, "socrates": 0.015449411226026707, "ibn_sina": 0.010618214699281045, "darwin": 0.00826947392262038, "hypatia": 0.005683520768258513}, "confidence": 0.8617486463055561, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 170.78517000118154, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "0479b973affcdcbaddd3c7ab9b3104ed54e2f719ac405e19ef024dec2504ae60", "compiled_ids_sha256": "3668378c5efff7a76c8a6932e67084fbfebd20b93955db9a4044351891f6a111", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 14.0625, 15.375, 15.0, 14.75, 14.375], "runtime_probabilities": [0.9558212161064148, 0.004158152732998133, 0.015449410304427147, 0.010618215426802635, 0.008269473910331726, 0.005683520343154669], "decide_seconds": 0.1666834910029138, "probabilities_f64_minus_runtime_max_abs": 1.0655897830780248e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.011479929384381488, "ibn_sina": 0.03536067191416106, "hypatia": 0.012220321218442406, "einstein": 0.9119637483665556, "sagan": 0.022830565714878237, "darwin": 0.006144763401581168}, "confidence": 0.7628512170817618, "choice": "einstein"}}, "usage": {"input_tokens": 345, "output_tokens": 1}, "latency_ms": 170.529652001278, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 345, "compiled_prompt_sha256": "c3c12676b722efd1aae12c93922ffc374ad0ba0d69b50bbaf70f9d7e70b298bf", "compiled_ids_sha256": "7d641b27474de1924e40dbf08e5dc53ed3bd65542e5474966c82971381d762dc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 16.25, 15.1875, 19.5, 15.8125, 14.5], "runtime_probabilities": [0.011479929089546204, 0.0353606715798378, 0.012220321223139763, 0.911963701248169, 0.02283056639134884, 0.006144763436168432], "decide_seconds": 0.16650993300208938, "probabilities_f64_minus_runtime_max_abs": 4.711838663684631e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:23Z"} {"id": "c-einstein-01-main-03", "uid": "c028", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "prompt_sha256_by_order": ["046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3"], "request_sha256": "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "compiled_prompt_sha256_by_order": ["23475b0a8910eef4fdc6c4e53fc0b009b8dddcf3b82d3c5b95467ce8d3a7fa19", "d5c8bac853b0e1d1b21e4a6fbaef7a867d13dd471060dad8608a82330852625d", "fb7a1f72ac4b85d3efe417cd470684f4e9b8df82cdba1cd52662f6ed4e8a1694", "bcb45eeb2d4da70be6304eba0a32c51cca067e0a883fb64f92e5015e0ee15acb", "6d8188e5da56048091a1bc58aa71d86c1aa92b521c07c4e77e85b1f010de6599", "629ccc60ca658ad9707ac652d5203d19c28bfa07038925151a13f5bd43db6135"], "compiled_tokens_by_order": [358, 358, 358, 358, 358, 358], "choice": "einstein", "probs": {"darwin": 0.004227233106507448, "einstein": 0.9717005366229027, "hypatia": 0.0024086048202218886, "ibn_sina": 0.001996803530153614, "sagan": 0.010140605061040503, "socrates": 0.009526216859173892}, "confidence": 0.9057768025618553, "seconds": 0.17420638550053505, "seconds_all_orders": 1.0452383130032104, "seconds_by_order": [0.17121987599966815, 0.17679036399931647, 0.1714208650009823, 0.17626668800221523, 0.17853632100013783, 0.17100419900089037], "argmax_stable": true, "spread": {"darwin": 0.008342257329758473, "einstein": 0.027869721034562533, "hypatia": 0.012102188221001972, "ibn_sina": 0.007505618768739085, "sagan": 0.029573624943574073, "socrates": 0.019349726372163093}, "spread_2dp": {"darwin": 0.01, "einstein": 0.030000000000000027, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.03, "socrates": 0.019999999999999997}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.004227233106507448, "einstein": 0.9717005366229027, "hypatia": 0.0024086048202218886, "ibn_sina": 0.001996803530153614, "sagan": 0.010140605061040503, "socrates": 0.009526216859173892}, {"ibn_sina": 0.009502422298892699, "hypatia": 0.0054143170192491935, "sagan": 0.0027224895704239934, "socrates": 0.0061352249547756135, "darwin": 0.006952120666720975, "einstein": 0.9692734254899374}, {"hypatia": 0.006859181597186524, "darwin": 0.003671455340181228, "sagan": 0.003240047965687605, "ibn_sina": 0.0044286204907247936, "einstein": 0.9563157432792813, "socrates": 0.025484951326938707}, {"ibn_sina": 0.008801240876246999, "einstein": 0.9556509393826212, "darwin": 0.0023688218461279226, "sagan": 0.004157411812556102, "hypatia": 0.01451079304122386, "socrates": 0.01451079304122386}, {"einstein": 0.9641806205460508, "sagan": 0.006915592486363533, "socrates": 0.008341796849720143, "ibn_sina": 0.004465042098196416, "darwin": 0.010711079175886396, "hypatia": 0.005385868843782643}, {"socrates": 0.009252991728206454, "ibn_sina": 0.006769633362869839, "hypatia": 0.0031977483749393884, "einstein": 0.9438308155883401, "sagan": 0.03229611451399807, "darwin": 0.004652696431646116}], "prompt_tokens": 358, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.004227233106507448, "einstein": 0.9717005366229027, "hypatia": 0.0024086048202218886, "ibn_sina": 0.001996803530153614, "sagan": 0.010140605061040503, "socrates": 0.009526216859173892}, "confidence": 0.9057768025618553, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 171.1919460030913, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "23475b0a8910eef4fdc6c4e53fc0b009b8dddcf3b82d3c5b95467ce8d3a7fa19", "compiled_ids_sha256": "74797ebb675858e67a4e4dc5542e122d2f6918fddd42e509d7955ac569f7cd1a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0625, 19.5, 13.5, 13.3125, 14.9375, 14.875], "runtime_probabilities": [0.0042272331193089485, 0.9717005491256714, 0.002408604836091399, 0.001996803330257535, 0.01014060527086258, 0.009526217356324196], "decide_seconds": 0.16686859000037657, "probabilities_f64_minus_runtime_max_abs": 1.2502768709232726e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.009502422298892699, "hypatia": 0.0054143170192491935, "sagan": 0.0027224895704239934, "socrates": 0.0061352249547756135, "darwin": 0.006952120666720975, "einstein": 0.9692734254899374}, "confidence": 0.8969592134219824, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 176.76563399800216, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "d5c8bac853b0e1d1b21e4a6fbaef7a867d13dd471060dad8608a82330852625d", "compiled_ids_sha256": "286586acffe27ea07f1ab127988e1ced1e45101be3885588aeadb470407e222a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.1875, 13.5, 14.3125, 14.4375, 19.375], "runtime_probabilities": [0.009502422995865345, 0.005414317362010479, 0.002722489880397916, 0.0061352248303592205, 0.006952120456844568, 0.9692734479904175], "decide_seconds": 0.1726208250001946, "probabilities_f64_minus_runtime_max_abs": 2.2500480034715054e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.006859181597186524, "darwin": 0.003671455340181228, "sagan": 0.003240047965687605, "ibn_sina": 0.0044286204907247936, "einstein": 0.9563157432792813, "socrates": 0.025484951326938707}, "confidence": 0.8696413301881117, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 171.3953649996256, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "fb7a1f72ac4b85d3efe417cd470684f4e9b8df82cdba1cd52662f6ed4e8a1694", "compiled_ids_sha256": "638ec1cf3b2e2e6011bf84fb8103217a4c11dc15a7641786a107d245e93cb377", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 14.1875, 14.0625, 14.375, 19.75, 16.125], "runtime_probabilities": [0.006859181448817253, 0.0036714556626975536, 0.003240048186853528, 0.004428620915859938, 0.956315815448761, 0.025484953075647354], "decide_seconds": 0.16706948799765087, "probabilities_f64_minus_runtime_max_abs": 7.21694797256589e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.008801240876246999, "einstein": 0.9556509393826212, "darwin": 0.0023688218461279226, "sagan": 0.004157411812556102, "hypatia": 0.01451079304122386, "socrates": 0.01451079304122386}, "confidence": 0.8632823606535371, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 176.23881799954688, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "bcb45eeb2d4da70be6304eba0a32c51cca067e0a883fb64f92e5015e0ee15acb", "compiled_ids_sha256": "e8f880dbee5dcd76cf1f2c023ebcf933b782730bb236941ce9e270417bfb4055", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 19.5, 13.5, 14.0625, 15.3125, 15.3125], "runtime_probabilities": [0.008801241405308247, 0.9556509852409363, 0.0023688222281634808, 0.004157412331551313, 0.01451079361140728, 0.01451079361140728], "decide_seconds": 0.1721220200015523, "probabilities_f64_minus_runtime_max_abs": 4.585831503423066e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9641806205460508, "sagan": 0.006915592486363533, "socrates": 0.008341796849720143, "ibn_sina": 0.004465042098196416, "darwin": 0.010711079175886396, "hypatia": 0.005385868843782643}, "confidence": 0.8825821406512053, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 178.51149099806207, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "6d8188e5da56048091a1bc58aa71d86c1aa92b521c07c4e77e85b1f010de6599", "compiled_ids_sha256": "4f28680994a18df493114096b811feeca2e7fccba6ef45f94247bb442e92e9e2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 14.5625, 14.75, 14.125, 15.0, 14.3125], "runtime_probabilities": [0.9641805291175842, 0.006915591657161713, 0.008341796696186066, 0.004465042147785425, 0.010711079463362694, 0.005385868716984987], "decide_seconds": 0.1743169730034424, "probabilities_f64_minus_runtime_max_abs": 9.142846657539394e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.009252991728206454, "ibn_sina": 0.006769633362869839, "hypatia": 0.0031977483749393884, "einstein": 0.9438308155883401, "sagan": 0.03229611451399807, "darwin": 0.004652696431646116}, "confidence": 0.8404180158013584, "choice": "einstein"}}, "usage": {"input_tokens": 358, "output_tokens": 1}, "latency_ms": 170.9794089983916, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 358, "compiled_prompt_sha256": "629ccc60ca658ad9707ac652d5203d19c28bfa07038925151a13f5bd43db6135", "compiled_ids_sha256": "a478eb11b273267fe901ecfa66d8fd870201a40c4c2e3b7ed607f04aab81aee8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 14.5625, 13.8125, 19.5, 16.125, 14.1875], "runtime_probabilities": [0.009252992458641529, 0.006769632920622826, 0.0031977484468370676, 0.9438308477401733, 0.032296113669872284, 0.0046526966616511345], "decide_seconds": 0.16682468999715638, "probabilities_f64_minus_runtime_max_abs": 3.215183319582593e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:24Z"} {"id": "c-einstein-01-main-03", "uid": "c029", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "prompt_sha256_by_order": ["1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979"], "request_sha256": "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "compiled_prompt_sha256_by_order": ["1f30f52d13f815504e7bf1165146b0c24247b90d1788ddb8d1a5b6473d203fe6", "fad5040fb67c25ac673e25a87b21d5f47d86e87e5aeec877e9176f428a184e3d", "a580632c4fd0505a5be8347d3492aeadeb88804466f558c10730641519263bba", "3cc8ae1c56a9863058534f4787fd929eef7ef7a37778b4f170d6952471538abb", "f673245e2835790db6990318e8f439fe617203b5a5956efef21eed45eea0ed61", "d7867483a63e187de51f957a47f9381866e9e5d3efd5b2458af608eae8da9eba"], "compiled_tokens_by_order": [402, 402, 402, 402, 402, 402], "choice": "einstein", "probs": {"darwin": 0.001237839626464866, "einstein": 0.993136009988375, "hypatia": 0.0007052997589310219, "ibn_sina": 0.0007052997589310219, "sagan": 0.0033647969631659777, "socrates": 0.0008507539041319242}, "confidence": 0.9717940451288577, "seconds": 0.1804556140001902, "seconds_all_orders": 1.0827336840011412, "seconds_by_order": [0.17728106999857118, 0.1807996429997729, 0.18593408400192857, 0.17720053100129007, 0.1777419860009104, 0.1837763699986681], "argmax_stable": true, "spread": {"darwin": 0.0009070749254783424, "einstein": 0.0044597898804972225, "hypatia": 0.0010300283620485492, "ibn_sina": 0.0007870483300454697, "sagan": 0.004101945974821711, "socrates": 0.0009615338968781114}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.01, "socrates": 0.0}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.001237839626464866, "einstein": 0.993136009988375, "hypatia": 0.0007052997589310219, "ibn_sina": 0.0007052997589310219, "sagan": 0.0033647969631659777, "socrates": 0.0008507539041319242}, {"ibn_sina": 0.0014923480889764917, "hypatia": 0.0015885962714921322, "sagan": 0.0014019312890492437, "socrates": 0.0014019312890492437, "darwin": 0.0014923480889764917, "einstein": 0.9926228449724563}, {"hypatia": 0.0014947826669825207, "darwin": 0.000965104226267219, "sagan": 0.0010936063610429555, "ibn_sina": 0.000965104226267219, "einstein": 0.9942421841631603, "socrates": 0.0012392183562798661}, {"ibn_sina": 0.0013189334339654033, "einstein": 0.9940856852587777, "darwin": 0.0005852731634981492, "sagan": 0.0016935440521308207, "hypatia": 0.0016935440521308207, "socrates": 0.0006230200394971063}, {"einstein": 0.9945568111077454, "sagan": 0.0014952556894621766, "socrates": 0.0011645063018451167, "ibn_sina": 0.0008003526963426723, "darwin": 0.0013195585145223646, "hypatia": 0.0006635156900822714}, {"socrates": 0.0015845539363752176, "ibn_sina": 0.0013136413525824287, "hypatia": 0.0009610805443799108, "einstein": 0.9900970212272482, "sagan": 0.005195552335864667, "darwin": 0.000848150603549588}], "prompt_tokens": 402, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.001237839626464866, "einstein": 0.993136009988375, "hypatia": 0.0007052997589310219, "ibn_sina": 0.0007052997589310219, "sagan": 0.0033647969631659777, "socrates": 0.0008507539041319242}, "confidence": 0.9717940451288577, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 177.25081000025966, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "1f30f52d13f815504e7bf1165146b0c24247b90d1788ddb8d1a5b6473d203fe6", "compiled_ids_sha256": "8e661bcb0784f987ceba2e82943447a2e43cffd2e0348e8aba9d9b56cec2f1d5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.9375, 19.625, 12.375, 12.375, 13.9375, 12.5625], "runtime_probabilities": [0.0012378396932035685, 0.9931360483169556, 0.0007052997825667262, 0.0007052997825667262, 0.003364797215908766, 0.0008507539168931544], "decide_seconds": 0.1724261880008271, "probabilities_f64_minus_runtime_max_abs": 3.8328580531654666e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0014923480889764917, "hypatia": 0.0015885962714921322, "sagan": 0.0014019312890492437, "socrates": 0.0014019312890492437, "darwin": 0.0014923480889764917, "einstein": 0.9926228449724563}, "confidence": 0.969062832604896, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 180.77390299731633, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "fad5040fb67c25ac673e25a87b21d5f47d86e87e5aeec877e9176f428a184e3d", "compiled_ids_sha256": "c34c9aa40ca503f588b1c0e1f62107facfb1588e5d86a3421a54bf52b2d80bc8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.875, 12.9375, 12.8125, 12.8125, 12.875, 19.375], "runtime_probabilities": [0.0014923480339348316, 0.001588596380315721, 0.0014019312802702188, 0.0014019312802702188, 0.0014923480339348316, 0.9926228523254395], "decide_seconds": 0.17616377900048974, "probabilities_f64_minus_runtime_max_abs": 7.35298311127508e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0014947826669825207, "darwin": 0.000965104226267219, "sagan": 0.0010936063610429555, "ibn_sina": 0.000965104226267219, "einstein": 0.9942421841631603, "socrates": 0.0012392183562798661}, "confidence": 0.9750977565368547, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 185.90763400061405, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "a580632c4fd0505a5be8347d3492aeadeb88804466f558c10730641519263bba", "compiled_ids_sha256": "686f4e052bae09a65511588e158762eed8dc752bf7ce44cfec785dd6221a492a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 12.8125, 12.9375, 12.8125, 19.75, 13.0625], "runtime_probabilities": [0.0014947826275601983, 0.0009651041473262012, 0.0010936063481494784, 0.0009651041473262012, 0.9942421317100525, 0.0012392183998599648], "decide_seconds": 0.1812303000006068, "probabilities_f64_minus_runtime_max_abs": 5.2453107857175496e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0013189334339654033, "einstein": 0.9940856852587777, "darwin": 0.0005852731634981492, "sagan": 0.0016935440521308207, "hypatia": 0.0016935440521308207, "socrates": 0.0006230200394971063}, "confidence": 0.9747676867609141, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 177.17382100090617, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "3cc8ae1c56a9863058534f4787fd929eef7ef7a37778b4f170d6952471538abb", "compiled_ids_sha256": "66e0ef5d15d470cf399bedf004fc077b33cf6bb3d11369cf2a24cead667e7e70", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.0, 19.625, 12.1875, 13.25, 13.25, 12.25], "runtime_probabilities": [0.001318933442234993, 0.9940856695175171, 0.000585273141041398, 0.0016935442108660936, 0.0016935442108660936, 0.0006230199942365289], "decide_seconds": 0.17234913799984497, "probabilities_f64_minus_runtime_max_abs": 1.5741260628310272e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9945568111077454, "sagan": 0.0014952556894621766, "socrates": 0.0011645063018451167, "ibn_sina": 0.0008003526963426723, "darwin": 0.0013195585145223646, "hypatia": 0.0006635156900822714}, "confidence": 0.9763728425200421, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 177.7162859980308, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "f673245e2835790db6990318e8f439fe617203b5a5956efef21eed45eea0ed61", "compiled_ids_sha256": "c02af61b97cc7e31a8f7047b29fa231950113da8e17d708229ee8d4e37624dab", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 13.125, 12.875, 12.5, 13.0, 12.3125], "runtime_probabilities": [0.9945567846298218, 0.0014952556230127811, 0.0011645061895251274, 0.0008003526600077748, 0.0013195584760978818, 0.0006635156460106373], "decide_seconds": 0.17311555199921713, "probabilities_f64_minus_runtime_max_abs": 2.647792363141832e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0015845539363752176, "ibn_sina": 0.0013136413525824287, "hypatia": 0.0009610805443799108, "einstein": 0.9900970212272482, "sagan": 0.005195552335864667, "darwin": 0.000848150603549588}, "confidence": 0.961607576127911, "choice": "einstein"}}, "usage": {"input_tokens": 402, "output_tokens": 1}, "latency_ms": 183.75055999786127, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 402, "compiled_prompt_sha256": "d7867483a63e187de51f957a47f9381866e9e5d3efd5b2458af608eae8da9eba", "compiled_ids_sha256": "8ad8767723780577622e71e68ef170e49ca04750277a35575b3d6c85464e7a7a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.0625, 12.875, 12.5625, 19.5, 14.25, 12.4375], "runtime_probabilities": [0.0015845539746806026, 0.0013136413181200624, 0.0009610804845578969, 0.9900970458984375, 0.005195552483201027, 0.0008481505792587996], "decide_seconds": 0.1789699369983282, "probabilities_f64_minus_runtime_max_abs": 2.4671189313707487e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:25Z"} {"id": "c-einstein-01-starter-01", "uid": "c030", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "prompt_sha256_by_order": ["08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa"], "request_sha256": "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "compiled_prompt_sha256_by_order": ["53cefe6d23d0ac7f3c46766dc20453dc7cb5499f889b97160ea8df183525bf2a", "fe7273a07d6d72bbbd5831f96a112ff514ec70058bfdca6741d13387e4ab5ef0", "15a6f6b712c4636d2c7ce9542482f92a3e9b34461b62a779ae89a82341c9c585", "ead9536380dfdb2c9e943e386ca4e01bede87b5bd9e36e3ea213aaf5177298bc", "37131f7cc41a227893b98bf82977596196305071846c5d2130db0a9fcc58a512", "e6ed2d8a7d165361f474e7cd117bd802f4b52681f976b8a960853337c4852e81"], "compiled_tokens_by_order": [305, 305, 305, 305, 305, 305], "choice": "einstein", "probs": {"darwin": 0.0035535813341595595, "einstein": 0.9853096147611875, "hypatia": 0.0010837810741319224, "ibn_sina": 0.0013072895153678871, "sagan": 0.007067140350451347, "socrates": 0.001678592964701767}, "confidence": 0.9461828770031024, "seconds": 0.16801966933295867, "seconds_all_orders": 1.008118015997752, "seconds_by_order": [0.17038346299887053, 0.1727556240002741, 0.16856872700009262, 0.16570622899962473, 0.16534499200133723, 0.16535898099755286], "argmax_stable": true, "spread": {"darwin": 0.003195744538184152, "einstein": 0.019833301882264398, "hypatia": 0.0010587436568031115, "ibn_sina": 0.00447568168103551, "sagan": 0.014029573018885745, "socrates": 0.0026472172458758855}, "spread_2dp": {"darwin": 0.0, "einstein": 0.020000000000000018, "hypatia": 0.0, "ibn_sina": 0.01, "sagan": 0.02, "socrates": 0.0}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0035535813341595595, "einstein": 0.9853096147611875, "hypatia": 0.0010837810741319224, "ibn_sina": 0.0013072895153678871, "sagan": 0.007067140350451347, "socrates": 0.001678592964701767}, {"ibn_sina": 0.0017975647766613242, "hypatia": 0.0015863453475987888, "sagan": 0.0016886558324490362, "socrates": 0.0017975647766613242, "darwin": 0.0019134977443018993, "einstein": 0.9912163715223277}, {"hypatia": 0.001902432966181123, "darwin": 0.00428719886963037, "sagan": 0.005171348961831588, "ibn_sina": 0.0015771723243504599, "einstein": 0.985484674553656, "socrates": 0.0015771723243504599}, {"ibn_sina": 0.002310608837803619, "einstein": 0.9922856862102738, "darwin": 0.0010914543314462179, "sagan": 0.0019155620072641738, "hypatia": 0.0014918411912791707, "socrates": 0.0009048474219330364}, {"einstein": 0.9848890852172868, "sagan": 0.005856364572718954, "socrates": 0.003552064667808922, "ibn_sina": 0.0013067315649983712, "darwin": 0.003552064667808922, "hypatia": 0.0008436893093780115}, {"socrates": 0.0032947197530866904, "ibn_sina": 0.005782413246033881, "hypatia": 0.001290230793631942, "einstein": 0.9724523843280094, "sagan": 0.01571822885133478, "darwin": 0.0014620230279032186}], "prompt_tokens": 305, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0035535813341595595, "einstein": 0.9853096147611875, "hypatia": 0.0010837810741319224, "ibn_sina": 0.0013072895153678871, "sagan": 0.007067140350451347, "socrates": 0.001678592964701767}, "confidence": 0.9461828770031024, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 170.35596300047473, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "53cefe6d23d0ac7f3c46766dc20453dc7cb5499f889b97160ea8df183525bf2a", "compiled_ids_sha256": "f3e5ac56a91af44389f871a8caca89d7e9b0d855b71d7a5f4fe583f155b126f2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 19.375, 12.5625, 12.75, 14.4375, 13.0], "runtime_probabilities": [0.0035535814240574837, 0.9853096604347229, 0.001083781011402607, 0.001307289581745863, 0.007067140191793442, 0.0016785931074991822], "decide_seconds": 0.16646999300064635, "probabilities_f64_minus_runtime_max_abs": 4.5673535398194076e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0017975647766613242, "hypatia": 0.0015863453475987888, "sagan": 0.0016886558324490362, "socrates": 0.0017975647766613242, "darwin": 0.0019134977443018993, "einstein": 0.9912163715223277}, "confidence": 0.9640279131468187, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 172.73360499893897, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "fe7273a07d6d72bbbd5831f96a112ff514ec70058bfdca6741d13387e4ab5ef0", "compiled_ids_sha256": "eb66ac5f446e0c30890d36c02b689c2df74bf26f69982ae2248faa56edd4ddbb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.0625, 12.9375, 13.0, 13.0625, 13.125, 19.375], "runtime_probabilities": [0.001797564560547471, 0.0015863452572375536, 0.0016886558150872588, 0.001797564560547471, 0.0019134975736960769, 0.9912163019180298], "decide_seconds": 0.16906234300040524, "probabilities_f64_minus_runtime_max_abs": 6.960429788094302e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.001902432966181123, "darwin": 0.00428719886963037, "sagan": 0.005171348961831588, "ibn_sina": 0.0015771723243504599, "einstein": 0.985484674553656, "socrates": 0.0015771723243504599}, "confidence": 0.9457074285171964, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 168.54592700110516, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "15a6f6b712c4636d2c7ce9542482f92a3e9b34461b62a779ae89a82341c9c585", "compiled_ids_sha256": "788547cc217fe137c8f0e6258247a249180ce82cf2d36c66eef85ce4af7a6301", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 14.0625, 14.25, 13.0625, 19.5, 13.0625], "runtime_probabilities": [0.0019024329958483577, 0.004287198651582003, 0.0051713488064706326, 0.0015771723119542003, 0.9854846596717834, 0.0015771723119542003], "decide_seconds": 0.1648628650000319, "probabilities_f64_minus_runtime_max_abs": 1.4881872512084726e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.002310608837803619, "einstein": 0.9922856862102738, "darwin": 0.0010914543314462179, "sagan": 0.0019155620072641738, "hypatia": 0.0014918411912791707, "socrates": 0.0009048474219330364}, "confidence": 0.9680811040969768, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 165.68282899970654, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "ead9536380dfdb2c9e943e386ca4e01bede87b5bd9e36e3ea213aaf5177298bc", "compiled_ids_sha256": "3b657d2333b7c8103126966e114c0abd541610cfa1782d2d1c3ab51889d46823", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 19.375, 12.5625, 13.125, 12.875, 12.375], "runtime_probabilities": [0.0023106089793145657, 0.9922857284545898, 0.0010914544109255075, 0.0019155620830133557, 0.001491841278038919, 0.0009048474603332579], "decide_seconds": 0.16185382799812942, "probabilities_f64_minus_runtime_max_abs": 4.2244316045270125e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9848890852172868, "sagan": 0.005856364572718954, "socrates": 0.003552064667808922, "ibn_sina": 0.0013067315649983712, "darwin": 0.003552064667808922, "hypatia": 0.0008436893093780115}, "confidence": 0.9442913527074647, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 165.3219319996424, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "37131f7cc41a227893b98bf82977596196305071846c5d2130db0a9fcc58a512", "compiled_ids_sha256": "fec24d9eede9f24754906876e38c33cb9abdb43055f366e6aecd3d7ce955483a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 14.5, 14.0, 13.0, 14.0, 12.5625], "runtime_probabilities": [0.9848891496658325, 0.0058563644997775555, 0.003552064998075366, 0.0013067316031083465, 0.003552064998075366, 0.0008436893112957478], "decide_seconds": 0.16167863000009675, "probabilities_f64_minus_runtime_max_abs": 6.444854572240644e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0032947197530866904, "ibn_sina": 0.005782413246033881, "hypatia": 0.001290230793631942, "einstein": 0.9724523843280094, "sagan": 0.01571822885133478, "darwin": 0.0014620230279032186}, "confidence": 0.9111508108452018, "choice": "einstein"}}, "usage": {"input_tokens": 305, "output_tokens": 1}, "latency_ms": 165.33650199926342, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 305, "compiled_prompt_sha256": "e6ed2d8a7d165361f474e7cd117bd802f4b52681f976b8a960853337c4852e81", "compiled_ids_sha256": "1fcf1f17b481dfc69aef5d0a87736aa7724c202eab5cc63d8fc32403a2b73ce1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.6875, 14.25, 12.75, 19.375, 15.25, 12.875], "runtime_probabilities": [0.003294719848781824, 0.005782413296401501, 0.001290230778977275, 0.9724524021148682, 0.015718229115009308, 0.0014620230067521334], "decide_seconds": 0.16165123999962816, "probabilities_f64_minus_runtime_max_abs": 1.7786858763635394e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:26Z"} {"id": "c-einstein-01-starter-02", "uid": "c031", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "prompt_sha256_by_order": ["06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7"], "request_sha256": "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "compiled_prompt_sha256_by_order": ["2c70462d4e85f4e64653498d9e754bf182bcbee31c8f1c5ece9f76079ffaacbf", "6ddbb2d9bc5a75bbb81da84886f84506b5ce64e9a7694668a9929a118db819ff", "0ae9b2b659ab144b74d101b700b2ff419a77540d4adc4d863325bc788325f9fa", "67887bf11c2ae2b8bce633d7ccbc05227fb3b22aed8d550200baa7aa4f1a89cb", "0444e28a162e609cf2a8b55f72729e5a229ab85fded73476d06e85590d5f61ef", "66e1cca3e42e537271045a99ce2635505d964d651ca26184d9f91ebefb77aedf"], "compiled_tokens_by_order": [267, 267, 267, 267, 267, 267], "choice": "sagan", "probs": {"darwin": 0.015779902992222167, "einstein": 0.02295962337719132, "hypatia": 0.06643587308395195, "ibn_sina": 0.010188271114103326, "sagan": 0.809354622619307, "socrates": 0.07528170681322416}, "confidence": 0.5842619655598762, "seconds": 0.16787968083372107, "seconds_all_orders": 1.0072780850023264, "seconds_by_order": [0.16488659500100766, 0.1722163090016693, 0.1675115959988034, 0.1669847689991002, 0.16471471700060647, 0.1709640990011394], "argmax_stable": false, "spread": {"darwin": 0.043331860788197035, "einstein": 0.5187948932837777, "hypatia": 0.21235087823793594, "ibn_sina": 0.022456765239039812, "sagan": 0.6463873637595328, "socrates": 0.18871891447441932}, "spread_2dp": {"darwin": 0.04, "einstein": 0.52, "hypatia": 0.22, "ibn_sina": 0.03, "sagan": 0.65, "socrates": 0.19}, "choices_by_order": ["sagan", "einstein", "einstein", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.015779902992222167, "einstein": 0.02295962337719132, "hypatia": 0.06643587308395195, "ibn_sina": 0.010188271114103326, "sagan": 0.809354622619307, "socrates": 0.07528170681322416}, {"ibn_sina": 0.023247993007621804, "hypatia": 0.023247993007621804, "sagan": 0.2832185344138292, "socrates": 0.09194759599906534, "darwin": 0.04921600158334671, "einstein": 0.5291218819885152}, {"hypatia": 0.03977202924186095, "darwin": 0.009446684962615325, "sagan": 0.2938777552364208, "ibn_sina": 0.027334889293692065, "einstein": 0.4275896108162614, "socrates": 0.2019790304491495}, {"ibn_sina": 0.012464243136345829, "einstein": 0.013268117753245098, "darwin": 0.013268117753245098, "sagan": 0.7711354168096509, "hypatia": 0.13400324231331573, "socrates": 0.05586086223419734}, {"einstein": 0.14289793880411994, "sagan": 0.5651723115175038, "socrates": 0.028138272513145664, "ibn_sina": 0.006278497252158352, "darwin": 0.021914108667514427, "hypatia": 0.23559887124555776}, {"socrates": 0.013260115974730178, "ibn_sina": 0.004878124054652253, "hypatia": 0.03604473229736854, "einstein": 0.010326988704737507, "sagan": 0.9296058981733619, "darwin": 0.005884140795149673}], "prompt_tokens": 267, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.015779902992222167, "einstein": 0.02295962337719132, "hypatia": 0.06643587308395195, "ibn_sina": 0.010188271114103326, "sagan": 0.809354622619307, "socrates": 0.07528170681322416}, "confidence": 0.5842619655598762, "choice": "sagan"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.85576500053867, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "2c70462d4e85f4e64653498d9e754bf182bcbee31c8f1c5ece9f76079ffaacbf", "compiled_ids_sha256": "fc5d36802beb2fda14c96e74d28c496080c202d461eda3e170bc4d68d9dc027a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.6875, 16.75, 14.875, 19.25, 16.875], "runtime_probabilities": [0.015779905021190643, 0.022959625348448753, 0.06643587350845337, 0.010188271291553974, 0.8093546628952026, 0.07528170943260193], "decide_seconds": 0.1613752410012239, "probabilities_f64_minus_runtime_max_abs": 4.027589561861333e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.023247993007621804, "hypatia": 0.023247993007621804, "sagan": 0.2832185344138292, "socrates": 0.09194759599906534, "darwin": 0.04921600158334671, "einstein": 0.5291218819885152}, "confidence": 0.3098156693289562, "choice": "einstein"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 172.1889590007777, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "6ddbb2d9bc5a75bbb81da84886f84506b5ce64e9a7694668a9929a118db819ff", "compiled_ids_sha256": "f1baa560653c22c4c9c528290a1b434b24533cb07f24573cdb7a6d20975dd839", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 15.625, 18.125, 17.0, 16.375, 18.75], "runtime_probabilities": [0.023247994482517242, 0.023247994482517242, 0.2832185626029968, 0.09194759279489517, 0.049216002225875854, 0.5291218757629395], "decide_seconds": 0.1688627039984567, "probabilities_f64_minus_runtime_max_abs": 2.8189167611269994e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.03977202924186095, "darwin": 0.009446684962615325, "sagan": 0.2938777552364208, "ibn_sina": 0.027334889293692065, "einstein": 0.4275896108162614, "socrates": 0.2019790304491495}, "confidence": 0.2650102585074441, "choice": "einstein"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 167.4887059998582, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "0ae9b2b659ab144b74d101b700b2ff419a77540d4adc4d863325bc788325f9fa", "compiled_ids_sha256": "f209506a405f4b1f64945fe9b066bb183c8cd249ff473cd5d22c9b503b415039", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 14.8125, 18.25, 15.875, 18.625, 17.875], "runtime_probabilities": [0.03977202996611595, 0.009446685202419758, 0.2938777804374695, 0.02733488939702511, 0.42758962512016296, 0.20197904109954834], "decide_seconds": 0.1640943310012517, "probabilities_f64_minus_runtime_max_abs": 2.5201048692302663e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.012464243136345829, "einstein": 0.013268117753245098, "darwin": 0.013268117753245098, "sagan": 0.7711354168096509, "hypatia": 0.13400324231331573, "socrates": 0.05586086223419734}, "confidence": 0.5533718706136872, "choice": "sagan"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 166.95960899960482, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "67887bf11c2ae2b8bce633d7ccbc05227fb3b22aed8d550200baa7aa4f1a89cb", "compiled_ids_sha256": "b7c191be89fc6f200713caae6420ff12ad72d98d1ef4b53aa41a03e772dc232f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 15.3125, 15.3125, 19.375, 17.625, 16.75], "runtime_probabilities": [0.012464242056012154, 0.013268116861581802, 0.013268116861581802, 0.7711353898048401, 0.13400322198867798, 0.055860862135887146], "decide_seconds": 0.16336365700044553, "probabilities_f64_minus_runtime_max_abs": 2.7004810831954273e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.14289793880411994, "sagan": 0.5651723115175038, "socrates": 0.028138272513145664, "ibn_sina": 0.006278497252158352, "darwin": 0.021914108667514427, "hypatia": 0.23559887124555776}, "confidence": 0.3541838485542236, "choice": "sagan"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.69080699971528, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "0444e28a162e609cf2a8b55f72729e5a229ab85fded73476d06e85590d5f61ef", "compiled_ids_sha256": "c9b5291c54ba363da61bbb5776949cccdab213fb6e3656843661917c82072d4d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.625, 19.0, 16.0, 14.5, 15.75, 18.125], "runtime_probabilities": [0.14289793372154236, 0.5651723146438599, 0.02813827246427536, 0.0062784976325929165, 0.021914107725024223, 0.2355988621711731], "decide_seconds": 0.16128784299871768, "probabilities_f64_minus_runtime_max_abs": 9.074384660356927e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.013260115974730178, "ibn_sina": 0.004878124054652253, "hypatia": 0.03604473229736854, "einstein": 0.010326988704737507, "sagan": 0.9296058981733619, "darwin": 0.005884140795149673}, "confidence": 0.8055735213741708, "choice": "sagan"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 170.94045900012134, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "66e1cca3e42e537271045a99ce2635505d964d651ca26184d9f91ebefb77aedf", "compiled_ids_sha256": "6aef6b34bf97625f7f131a5d5f75df2857a47756da3f3470ae57d51c8fde5754", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.0, 16.0, 14.75, 19.25, 14.1875], "runtime_probabilities": [0.013260114938020706, 0.004878123756498098, 0.036044731736183167, 0.010326988995075226, 0.9296058416366577, 0.0058841402642428875], "decide_seconds": 0.16755064400058473, "probabilities_f64_minus_runtime_max_abs": 5.653670420713297e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:27Z"} {"id": "c-einstein-01-starter-02", "uid": "c032", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "prompt_sha256_by_order": ["8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db"], "request_sha256": "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "compiled_prompt_sha256_by_order": ["370a32f6cf6865231ae43ab9a2c0130f1bd393f6ff9c93f7322c5eec6f69a7f6", "e52019829e848e766454e6a6ea48668b93f4bcca340e9c914614ef8f170050d4", "3b3121397ccf675eaf9eea7a8f22638372036d0a56b6ccaf93cff2422d317294", "4e63cd420f0e8ae84d5b86bdea3c661c9b72ffadd05ea27f762b7d6f36a6024c", "f60a7fd29e7b7e6df5f25948d1fc5cb160b82afd34a998f4d5847d53c378da07", "b076d723c9254fe45a7e5a59c0a520f7484fd0ab3cbd9891413450ae9e75c36d"], "compiled_tokens_by_order": [294, 294, 294, 294, 294, 294], "choice": "einstein", "probs": {"darwin": 0.022272198397103954, "einstein": 0.39478416270600575, "hypatia": 0.03240585745245952, "ibn_sina": 0.010520641578803948, "sagan": 0.39478416270600575, "socrates": 0.14523297715962116}, "confidence": 0.2979896992527806, "seconds": 0.16809409749960955, "seconds_all_orders": 1.0085645849976572, "seconds_by_order": [0.17158938400098123, 0.17041101299764705, 0.16758789400046226, 0.16571258900148678, 0.16524437299813144, 0.16801933199894847], "argmax_stable": false, "spread": {"darwin": 0.08356073122200271, "einstein": 0.5798786917566692, "hypatia": 0.06093071617226621, "ibn_sina": 0.018547443071340833, "sagan": 0.6762751034974284, "socrates": 0.35140714887954494}, "spread_2dp": {"darwin": 0.09000000000000001, "einstein": 0.58, "hypatia": 0.06, "ibn_sina": 0.019999999999999997, "sagan": 0.6799999999999999, "socrates": 0.35}, "choices_by_order": ["einstein", "einstein", "einstein", "sagan", "einstein", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.022272198397103954, "einstein": 0.39478416270600575, "hypatia": 0.03240585745245952, "ibn_sina": 0.010520641578803948, "sagan": 0.39478416270600575, "socrates": 0.14523297715962116}, {"ibn_sina": 0.02132806597010736, "hypatia": 0.024167864960934374, "sagan": 0.05116337052379095, "socrates": 0.1390764603775381, "darwin": 0.05797569416271858, "einstein": 0.7062885440049107}, {"hypatia": 0.021144239920144425, "darwin": 0.012047624748488237, "sagan": 0.08362709113201125, "ibn_sina": 0.02714974147400902, "einstein": 0.481240682414847, "socrates": 0.37479062031050014}, {"ibn_sina": 0.01720384539689564, "einstein": 0.19688758442351822, "darwin": 0.0342139319617072, "sagan": 0.4168110194954175, "hypatia": 0.08207495609241064, "socrates": 0.25280866263005086}, {"einstein": 0.5501881011162956, "sagan": 0.157631530145218, "socrates": 0.10833846066822428, "ibn_sina": 0.013773689601783202, "darwin": 0.09560835597049094, "hypatia": 0.07445986249798797}, {"socrates": 0.023383471430955198, "ibn_sina": 0.008602298402668187, "hypatia": 0.06766231809326821, "einstein": 0.12640985224824147, "sagan": 0.7274384740212193, "darwin": 0.046503585803647655}], "prompt_tokens": 294, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.022272198397103954, "einstein": 0.39478416270600575, "hypatia": 0.03240585745245952, "ibn_sina": 0.010520641578803948, "sagan": 0.39478416270600575, "socrates": 0.14523297715962116}, "confidence": 0.2979896992527806, "choice": "einstein"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 171.5628639976785, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "370a32f6cf6865231ae43ab9a2c0130f1bd393f6ff9c93f7322c5eec6f69a7f6", "compiled_ids_sha256": "45bd3725a496f6775e0e3d110492d7c95aeb7acbd0c5e55b8758c122228a0ab4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.875, 18.75, 16.25, 15.125, 18.75, 17.75], "runtime_probabilities": [0.022272199392318726, 0.394784152507782, 0.03240585699677467, 0.010520640760660172, 0.394784152507782, 0.14523299038410187], "decide_seconds": 0.16779295300148078, "probabilities_f64_minus_runtime_max_abs": 1.3224480704954189e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02132806597010736, "hypatia": 0.024167864960934374, "sagan": 0.05116337052379095, "socrates": 0.1390764603775381, "darwin": 0.05797569416271858, "einstein": 0.7062885440049107}, "confidence": 0.4367611579320573, "choice": "einstein"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 170.3876730025513, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "e52019829e848e766454e6a6ea48668b93f4bcca340e9c914614ef8f170050d4", "compiled_ids_sha256": "0b825118dfe92a780f2d6b3cb156be8e2c51d191e0e73f63c14f0973f77e565a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 15.75, 16.5, 17.5, 16.625, 19.125], "runtime_probabilities": [0.021328065544366837, 0.0241678636521101, 0.05116337165236473, 0.13907645642757416, 0.05797569453716278, 0.7062885165214539], "decide_seconds": 0.16663605099893175, "probabilities_f64_minus_runtime_max_abs": 2.748345684366882e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.021144239920144425, "darwin": 0.012047624748488237, "sagan": 0.08362709113201125, "ibn_sina": 0.02714974147400902, "einstein": 0.481240682414847, "socrates": 0.37479062031050014}, "confidence": 0.35259742615443934, "choice": "einstein"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 167.56500500196125, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "3b3121397ccf675eaf9eea7a8f22638372036d0a56b6ccaf93cff2422d317294", "compiled_ids_sha256": "943ef61d197eea000fa0604661bd388bd860b1b6552a1a21d94cb1c85af2426c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 15.4375, 17.375, 16.25, 19.125, 18.875], "runtime_probabilities": [0.021144239231944084, 0.01204762514680624, 0.08362708240747452, 0.02714974246919155, 0.4812406599521637, 0.37479060888290405], "decide_seconds": 0.16394829300043057, "probabilities_f64_minus_runtime_max_abs": 2.246268332450896e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01720384539689564, "einstein": 0.19688758442351822, "darwin": 0.0342139319617072, "sagan": 0.4168110194954175, "hypatia": 0.08207495609241064, "socrates": 0.25280866263005086}, "confidence": 0.20584415185678995, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 165.69020899987663, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "4e63cd420f0e8ae84d5b86bdea3c661c9b72ffadd05ea27f762b7d6f36a6024c", "compiled_ids_sha256": "8b018fb09a854ef86715772ca5aec50aa39589b6d733ca573546abe6e0bb32fd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 18.125, 16.375, 18.875, 17.25, 18.375], "runtime_probabilities": [0.01720384508371353, 0.19688758254051208, 0.03421393036842346, 0.4168109893798828, 0.08207494765520096, 0.25280866026878357], "decide_seconds": 0.16207862699957332, "probabilities_f64_minus_runtime_max_abs": 3.011553467713313e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.5501881011162956, "sagan": 0.157631530145218, "socrates": 0.10833846066822428, "ibn_sina": 0.013773689601783202, "darwin": 0.09560835597049094, "hypatia": 0.07445986249798797}, "confidence": 0.25346603727455885, "choice": "einstein"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 165.2202329969441, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "f60a7fd29e7b7e6df5f25948d1fc5cb160b82afd34a998f4d5847d53c378da07", "compiled_ids_sha256": "545c034c3c21ebbfc4845d601901f40ca7982940480a018d8a8d0660d31376a4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 18.125, 17.75, 15.6875, 17.625, 17.375], "runtime_probabilities": [0.5501881241798401, 0.1576315313577652, 0.10833845287561417, 0.013773689977824688, 0.09560834616422653, 0.07445986568927765], "decide_seconds": 0.16154275100052473, "probabilities_f64_minus_runtime_max_abs": 2.306354451864223e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.023383471430955198, "ibn_sina": 0.008602298402668187, "hypatia": 0.06766231809326821, "einstein": 0.12640985224824147, "sagan": 0.7274384740212193, "darwin": 0.046503585803647655}, "confidence": 0.47170400847131944, "choice": "sagan"}}, "usage": {"input_tokens": 294, "output_tokens": 1}, "latency_ms": 167.99706199890352, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 294, "compiled_prompt_sha256": "b076d723c9254fe45a7e5a59c0a520f7484fd0ab3cbd9891413450ae9e75c36d", "compiled_ids_sha256": "75b3c95af00e5cc43dd13c6f18112755b3bfdc54a67f447b092a77fde8f7f9c9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 14.9375, 17.0, 17.625, 19.375, 16.625], "runtime_probabilities": [0.023383473977446556, 0.00860229879617691, 0.0676623210310936, 0.1264098584651947, 0.7274385094642639, 0.04650358855724335], "decide_seconds": 0.16443037899807678, "probabilities_f64_minus_runtime_max_abs": 3.5443044610339314e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:28Z"} {"id": "c-einstein-01-starter-03", "uid": "c033", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "prompt_sha256_by_order": ["ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c"], "request_sha256": "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "compiled_prompt_sha256_by_order": ["32dea34ab78631d1128469f41a9a9b72979f369974ba4b1ff7cdc18dca902d44", "b2dc88d839a341d8e5dbb739cd7f24a26c813ab1a286765ea240e741b32f962d", "b429c3380c2f0253b5f360120a373ebdfca865dff32f4f83aa8149409771f11a", "32569ed4b1860f0ccaa5b47061cf534a8b0cec2159b6ea6dcf44768ba0300d0f", "d95158421940fd1c60b8d6778bc47021fd09a020413870cd075808a0690e9efb", "7e24416d18a1e23b747b58650780df0865e19627020d3b2ca042eb06a7df934c"], "compiled_tokens_by_order": [387, 387, 387, 387, 387, 387], "choice": "einstein", "probs": {"darwin": 0.008126018209746259, "einstein": 0.9392400008284197, "hypatia": 0.004928679185594272, "ibn_sina": 0.0031821944534147642, "sagan": 0.028362590430594536, "socrates": 0.0161605168922304}, "confidence": 0.826886200196703, "seconds": 0.17831456383343416, "seconds_all_orders": 1.069887383000605, "seconds_by_order": [0.17727336999814725, 0.1765068060012709, 0.18165849599972717, 0.17731606000234024, 0.1791232260002289, 0.1780094249988906], "argmax_stable": true, "spread": {"darwin": 0.010716826651515196, "einstein": 0.3536057513606232, "hypatia": 0.09009345135450125, "ibn_sina": 0.007729120747025316, "sagan": 0.10111662237508866, "socrates": 0.14824493685336096}, "spread_2dp": {"darwin": 0.01, "einstein": 0.35, "hypatia": 0.1, "ibn_sina": 0.01, "sagan": 0.1, "socrates": 0.15}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.008126018209746259, "einstein": 0.9392400008284197, "hypatia": 0.004928679185594272, "ibn_sina": 0.0031821944534147642, "sagan": 0.028362590430594536, "socrates": 0.0161605168922304}, {"ibn_sina": 0.007586568209213341, "hypatia": 0.005550451871930217, "sagan": 0.00807585982090999, "socrates": 0.036193492692954625, "darwin": 0.009151148063248692, "einstein": 0.9334424793417431}, {"hypatia": 0.005787024494263577, "darwin": 0.003097570997673909, "sagan": 0.006557557853534605, "ibn_sina": 0.0029098986582070407, "einstein": 0.973227877040982, "socrates": 0.008420070955338742}, {"ibn_sina": 0.010639019405232356, "einstein": 0.8996702600811752, "darwin": 0.005349634915596447, "sagan": 0.03488400164440688, "hypatia": 0.030785023400945, "socrates": 0.018672060552644032}, {"einstein": 0.6196221256803588, "sagan": 0.10767418022862327, "socrates": 0.1566650078086997, "ibn_sina": 0.007327319921000681, "darwin": 0.013689235821222112, "hypatia": 0.09502213054009552}, {"socrates": 0.015095139313728244, "ibn_sina": 0.0040628016298483146, "hypatia": 0.00807984583275935, "einstein": 0.8773208651728001, "sagan": 0.09246893888115705, "darwin": 0.002972409169706916}], "prompt_tokens": 387, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.008126018209746259, "einstein": 0.9392400008284197, "hypatia": 0.004928679185594272, "ibn_sina": 0.0031821944534147642, "sagan": 0.028362590430594536, "socrates": 0.0161605168922304}, "confidence": 0.826886200196703, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 177.24054000063916, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "32dea34ab78631d1128469f41a9a9b72979f369974ba4b1ff7cdc18dca902d44", "compiled_ids_sha256": "4844b40c337845fad28c4f6e260cd19b165cd0fb8800c232d4d2050f7333a62b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 19.375, 14.125, 13.6875, 15.875, 15.3125], "runtime_probabilities": [0.008126018568873405, 0.939240038394928, 0.004928679205477238, 0.0031821944285184145, 0.028362590819597244, 0.016160516068339348], "decide_seconds": 0.17260098599945195, "probabilities_f64_minus_runtime_max_abs": 3.756650823749652e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.007586568209213341, "hypatia": 0.005550451871930217, "sagan": 0.00807585982090999, "socrates": 0.036193492692954625, "darwin": 0.009151148063248692, "einstein": 0.9334424793417431}, "confidence": 0.8146260073290894, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 176.47993599894107, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "b2dc88d839a341d8e5dbb739cd7f24a26c813ab1a286765ea240e741b32f962d", "compiled_ids_sha256": "81c845b35e69d0dcc40433722e5747ef371bb7881b11bc6acb0d5dd0398ef11c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.25, 14.625, 16.125, 14.75, 19.375], "runtime_probabilities": [0.007586569059640169, 0.005550452042371035, 0.00807586032897234, 0.036193493753671646, 0.009151148609817028, 0.9334425330162048], "decide_seconds": 0.17200417999993078, "probabilities_f64_minus_runtime_max_abs": 5.3674461764074977e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.005787024494263577, "darwin": 0.003097570997673909, "sagan": 0.006557557853534605, "ibn_sina": 0.0029098986582070407, "einstein": 0.973227877040982, "socrates": 0.008420070955338742}, "confidence": 0.9083004875311577, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 181.63202600044315, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "b429c3380c2f0253b5f360120a373ebdfca865dff32f4f83aa8149409771f11a", "compiled_ids_sha256": "8d69cf61a4b78ca20ed2eeb182851cbb6fd5610240665d51b41f2bdd84be84ae", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 13.75, 14.5, 13.6875, 19.5, 14.75], "runtime_probabilities": [0.005787024740129709, 0.0030975714325904846, 0.006557558197528124, 0.002909898990765214, 0.9732279777526855, 0.008420071564614773], "decide_seconds": 0.17719656099870917, "probabilities_f64_minus_runtime_max_abs": 1.0071170353231906e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.010639019405232356, "einstein": 0.8996702600811752, "darwin": 0.005349634915596447, "sagan": 0.03488400164440688, "hypatia": 0.030785023400945, "socrates": 0.018672060552644032}, "confidence": 0.7376985080493241, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 177.28834000081406, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "32569ed4b1860f0ccaa5b47061cf534a8b0cec2159b6ea6dcf44768ba0300d0f", "compiled_ids_sha256": "8db1e8b99c418619af59672c37d9a945d44c227d8d1e9b74ae3dce7467be82c9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 19.25, 14.125, 16.0, 15.875, 15.375], "runtime_probabilities": [0.010639018379151821, 0.8996702432632446, 0.0053496346808969975, 0.034883998334407806, 0.03078502230346203, 0.01867206022143364], "decide_seconds": 0.17275659500228358, "probabilities_f64_minus_runtime_max_abs": 1.681793060370751e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.6196221256803588, "sagan": 0.10767418022862327, "socrates": 0.1566650078086997, "ibn_sina": 0.007327319921000681, "darwin": 0.013689235821222112, "hypatia": 0.09502213054009552}, "confidence": 0.3607616587240099, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 179.09493600018322, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "d95158421940fd1c60b8d6778bc47021fd09a020413870cd075808a0690e9efb", "compiled_ids_sha256": "d01b6c896e4280e721feef7812f08140bf588566002b821e2329a6254e37509f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 17.625, 18.0, 14.9375, 15.5625, 17.5], "runtime_probabilities": [0.6196221709251404, 0.10767418146133423, 0.15666502714157104, 0.0073273200541734695, 0.013689236715435982, 0.09502214193344116], "decide_seconds": 0.17447913200157927, "probabilities_f64_minus_runtime_max_abs": 4.524478158796086e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015095139313728244, "ibn_sina": 0.0040628016298483146, "hypatia": 0.00807984583275935, "einstein": 0.8773208651728001, "sagan": 0.09246893888115705, "darwin": 0.002972409169706916}, "confidence": 0.7338487536279175, "choice": "einstein"}}, "usage": {"input_tokens": 387, "output_tokens": 1}, "latency_ms": 177.98061500070617, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 387, "compiled_prompt_sha256": "7e24416d18a1e23b747b58650780df0865e19627020d3b2ca042eb06a7df934c", "compiled_ids_sha256": "58234122cb361c455f36bd355cee04afccae04df794d50603124bb2b2e97571b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.125, 14.8125, 19.5, 17.25, 13.8125], "runtime_probabilities": [0.015095138922333717, 0.004062802065163851, 0.00807984545826912, 0.8773208856582642, 0.09246893972158432, 0.002972409361973405], "decide_seconds": 0.17330601000139723, "probabilities_f64_minus_runtime_max_abs": 2.0485464102826256e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:29Z"} {"id": "c-einstein-01-starter-03", "uid": "c034", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "prompt_sha256_by_order": ["70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c"], "request_sha256": "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "compiled_prompt_sha256_by_order": ["3048ef259ba2c81faa325128892aa67888378d9abaef64bcc7e191dc46c7930d", "7d0d1eb128df1b1634aec02eca91a38321fbf48bd80a3b1ce481bfe00dea0618", "11faceff6b952748ccf699acff210388b55eaa906f0f34d2ae0b2e77649e452a", "ee56731ff21bb019336e14f79c19363ac8b016f5bf686ae6a3bd61984dd9a695", "b645aa52c104637afc161ca122c71a905a2d51cc15286d6d4fea893137dde86f", "6e1c2ab21ae1afd1555a09dc961bb6e14a89b5c4d6efdb6289f36bb9106295c6"], "compiled_tokens_by_order": [343, 343, 343, 343, 343, 343], "choice": "einstein", "probs": {"darwin": 0.0020364458425125532, "einstein": 0.990991596364364, "hypatia": 0.0011603318645583471, "ibn_sina": 0.0011603318645583471, "sagan": 0.0026148482214942, "socrates": 0.0020364458425125532}, "confidence": 0.9634770266015207, "seconds": 0.1730044830001134, "seconds_all_orders": 1.0380268980006804, "seconds_by_order": [0.17463744999986375, 0.17037589300161926, 0.17317321199880098, 0.1709160089994839, 0.17838861199925304, 0.17053572200165945], "argmax_stable": true, "spread": {"darwin": 0.0026561245485686715, "einstein": 0.004717261350233759, "hypatia": 0.0028773836531125994, "ibn_sina": 0.0017937239134100979, "sagan": 0.0025422446300548056, "socrates": 0.0017616872413612613}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0020364458425125532, "einstein": 0.990991596364364, "hypatia": 0.0011603318645583471, "ibn_sina": 0.0011603318645583471, "sagan": 0.0026148482214942, "socrates": 0.0020364458425125532}, {"ibn_sina": 0.0023064322706998247, "hypatia": 0.0016874218963508834, "sagan": 0.0010234731160072442, "socrates": 0.0024551843720292997, "darwin": 0.00203541933491375, "einstein": 0.9904920690099989}, {"hypatia": 0.002610416147967765, "darwin": 0.0013972550763902657, "sagan": 0.0010881833476432347, "ibn_sina": 0.001794111031681073, "einstein": 0.9893119013124437, "socrates": 0.0037981330838738146}, {"ibn_sina": 0.002954055777968445, "einstein": 0.987996411952131, "darwin": 0.0009009361100917187, "sagan": 0.001156824864169552, "hypatia": 0.004037715517670947, "socrates": 0.002954055777968445}, {"einstein": 0.9862743350141302, "sagan": 0.0022966109726787086, "socrates": 0.0035570606586603905, "ibn_sina": 0.002157466347935141, "darwin": 0.0035570606586603905, "hypatia": 0.002157466347935141}, {"socrates": 0.0029560838386980255, "ibn_sina": 0.0020316847295444146, "hypatia": 0.0016843257997547912, "einstein": 0.9886747054153047, "sagan": 0.0035657177460620497, "darwin": 0.0010874824706361616}], "prompt_tokens": 343, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0020364458425125532, "einstein": 0.990991596364364, "hypatia": 0.0011603318645583471, "ibn_sina": 0.0011603318645583471, "sagan": 0.0026148482214942, "socrates": 0.0020364458425125532}, "confidence": 0.9634770266015207, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 174.6095199996489, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "3048ef259ba2c81faa325128892aa67888378d9abaef64bcc7e191dc46c7930d", "compiled_ids_sha256": "dee894996ae3ae3d47634ad4330ad3215bf4766aff9df1233b15480c06ddd321", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.0625, 19.25, 12.5, 12.5, 13.3125, 13.0625], "runtime_probabilities": [0.002036445774137974, 0.9909915924072266, 0.0011603317689150572, 0.0011603317689150572, 0.00261484831571579, 0.002036445774137974], "decide_seconds": 0.17043735300103435, "probabilities_f64_minus_runtime_max_abs": 3.957137395538268e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0023064322706998247, "hypatia": 0.0016874218963508834, "sagan": 0.0010234731160072442, "socrates": 0.0024551843720292997, "darwin": 0.00203541933491375, "einstein": 0.9904920690099989}, "confidence": 0.9616828472897264, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.35042300267378, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "7d0d1eb128df1b1634aec02eca91a38321fbf48bd80a3b1ce481bfe00dea0618", "compiled_ids_sha256": "9c06a7708d2042ac30c7d558072e3af4c5b5e55520e466b3ffd8ae4667ccbd41", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 12.875, 12.375, 13.25, 13.0625, 19.25], "runtime_probabilities": [0.002306432230398059, 0.001687422045506537, 0.0010234731016680598, 0.002455184468999505, 0.002035419223830104, 0.990492045879364], "decide_seconds": 0.16626584499681485, "probabilities_f64_minus_runtime_max_abs": 2.313063485193112e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.002610416147967765, "darwin": 0.0013972550763902657, "sagan": 0.0010881833476432347, "ibn_sina": 0.001794111031681073, "einstein": 0.9893119013124437, "socrates": 0.0037981330838738146}, "confidence": 0.9579853487456818, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 173.1480720009131, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "11faceff6b952748ccf699acff210388b55eaa906f0f34d2ae0b2e77649e452a", "compiled_ids_sha256": "29aa305dd02ddde621698702f0ad4b9afa944197055976ac3b0a3b010b67016d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 12.9375, 12.6875, 13.1875, 19.5, 13.9375], "runtime_probabilities": [0.0026104161515831947, 0.0013972551096230745, 0.0010881832567974925, 0.0017941109836101532, 0.9893118739128113, 0.0037981332279741764], "decide_seconds": 0.16909062299964717, "probabilities_f64_minus_runtime_max_abs": 2.7399632451796663e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.002954055777968445, "einstein": 0.987996411952131, "darwin": 0.0009009361100917187, "sagan": 0.001156824864169552, "hypatia": 0.004037715517670947, "socrates": 0.002954055777968445}, "confidence": 0.9538221124770904, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.89201899943873, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "ee56731ff21bb019336e14f79c19363ac8b016f5bf686ae6a3bd61984dd9a695", "compiled_ids_sha256": "b4bc04d940bf7bd555bc38846260b794f9414ea8300db287ad490da47b828946", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 19.375, 12.375, 12.625, 13.875, 13.5625], "runtime_probabilities": [0.0029540560208261013, 0.987996518611908, 0.0009009361965581775, 0.0011568248737603426, 0.004037715960294008, 0.0029540560208261013], "decide_seconds": 0.16676294100034283, "probabilities_f64_minus_runtime_max_abs": 1.0665977701052043e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9862743350141302, "sagan": 0.0022966109726787086, "socrates": 0.0035570606586603905, "ibn_sina": 0.002157466347935141, "darwin": 0.0035570606586603905, "hypatia": 0.002157466347935141}, "confidence": 0.9474316538282446, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 178.36348199853092, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "b645aa52c104637afc161ca122c71a905a2d51cc15286d6d4fea893137dde86f", "compiled_ids_sha256": "6cc6d697b909a6695bb779df4e748c793556a899be9c8812c76e593c88709b55", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 13.4375, 13.875, 13.375, 13.875, 13.375], "runtime_probabilities": [0.9862744212150574, 0.0022966109681874514, 0.0035570610780268908, 0.002157466486096382, 0.0035570610780268908, 0.002157466486096382], "decide_seconds": 0.17413217400098802, "probabilities_f64_minus_runtime_max_abs": 8.620092717404049e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0029560838386980255, "ibn_sina": 0.0020316847295444146, "hypatia": 0.0016843257997547912, "einstein": 0.9886747054153047, "sagan": 0.0035657177460620497, "darwin": 0.0010874824706361616}, "confidence": 0.9557159230063854, "choice": "einstein"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.51160199844162, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "6e1c2ab21ae1afd1555a09dc961bb6e14a89b5c4d6efdb6289f36bb9106295c6", "compiled_ids_sha256": "e49b95bdfc653d05318070b9204906b2f7090777af85b775a6215ee2a0fcc56d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 13.1875, 13.0, 19.375, 13.75, 12.5625], "runtime_probabilities": [0.002956083742901683, 0.002031684620305896, 0.0016843258636072278, 0.9886747002601624, 0.0035657177213579416, 0.0010874824365600944], "decide_seconds": 0.16638788399723126, "probabilities_f64_minus_runtime_max_abs": 5.15514231302916e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:30Z"} {"id": "c-einstein-01-starter-03", "uid": "c035", "task": "kperm", "label": "einstein", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "prompt_sha256_by_order": ["36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156"], "request_sha256": "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "compiled_prompt_sha256_by_order": ["8f4e67e82afe236dfe30a2d5eeec24f6f2778b73186fa1f61532a001b7adee48", "56ba92fcf591d9e229f3f8ad5237c5869776100cf3a70c6bec64ce9064009b29", "c78dcb949364652ccb81e7ece0a7c58511b69ae2b059a4200cc368f7cd710b86", "8b89de27ed879d98c030e2f398520cd0febf51c68c5518567109e07953145424", "c0960b215a98d11febcac1885a2697564706bea112c92b9221476c4435d5d8b9", "af19bfbc7bf8d777b190a6366cb8b8ef37d3d9885df8ab53705093e1856fea10"], "compiled_tokens_by_order": [278, 278, 278, 278, 278, 278], "choice": "einstein", "probs": {"darwin": 0.0031313210112257576, "einstein": 0.9838318160834885, "hypatia": 0.002152123359468369, "ibn_sina": 0.001676075357620228, "sagan": 0.007056540828728887, "socrates": 0.002152123359468369}, "confidence": 0.9407311222901245, "seconds": 0.1696861253331008, "seconds_all_orders": 1.0181167519986047, "seconds_by_order": [0.17051374199945712, 0.1732276520015148, 0.1690066739975009, 0.16534364100152743, 0.17096688899982837, 0.1690581539987761], "argmax_stable": true, "spread": {"darwin": 0.00309776186513419, "einstein": 0.07563482007748967, "hypatia": 0.007564308779256349, "ibn_sina": 0.0028529755780140422, "sagan": 0.06874238645069457, "socrates": 0.0018147085619868763}, "spread_2dp": {"darwin": 0.0, "einstein": 0.06999999999999995, "hypatia": 0.01, "ibn_sina": 0.0, "sagan": 0.060000000000000005, "socrates": 0.0}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0031313210112257576, "einstein": 0.9838318160834885, "hypatia": 0.002152123359468369, "ibn_sina": 0.001676075357620228, "sagan": 0.007056540828728887, "socrates": 0.002152123359468369}, {"ibn_sina": 0.003960998142614711, "hypatia": 0.0037210133969521747, "sagan": 0.017751962083385725, "socrates": 0.0022569087104529497, "darwin": 0.003084828455212717, "einstein": 0.9692242892113817}, {"hypatia": 0.0036559613437314236, "darwin": 0.0019568950914558886, "sagan": 0.036923901561872675, "ibn_sina": 0.0019568950914558886, "einstein": 0.9522799723496722, "socrates": 0.0032263745618119964}, {"ibn_sina": 0.004522467111022192, "einstein": 0.9765809704830842, "darwin": 0.001379274540924254, "sagan": 0.005806962716686877, "hypatia": 0.009574062949164255, "socrates": 0.002136262199118169}, {"einstein": 0.9081969960059988, "sagan": 0.07454934916738144, "socrates": 0.003950970761105045, "ibn_sina": 0.003077019122640834, "darwin": 0.004477036406058444, "hypatia": 0.0057486285368153825}, {"socrates": 0.0031127681511596557, "ibn_sina": 0.00452905093563427, "hypatia": 0.0020097541699079058, "einstein": 0.9780026807291319, "sagan": 0.010206373836512359, "darwin": 0.002139372177654028}], "prompt_tokens": 278, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0031313210112257576, "einstein": 0.9838318160834885, "hypatia": 0.002152123359468369, "ibn_sina": 0.001676075357620228, "sagan": 0.007056540828728887, "socrates": 0.002152123359468369}, "confidence": 0.9407311222901245, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 170.4883320016961, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "8f4e67e82afe236dfe30a2d5eeec24f6f2778b73186fa1f61532a001b7adee48", "compiled_ids_sha256": "bc3a2e6b1255920114fcfe262b849bf8a7b90aa700f811ca5a13e3dea051bd85", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 19.75, 13.625, 13.375, 14.8125, 13.625], "runtime_probabilities": [0.0031313213985413313, 0.9838318824768066, 0.002152123488485813, 0.0016760756261646748, 0.007056540809571743, 0.002152123488485813], "decide_seconds": 0.16701706900130375, "probabilities_f64_minus_runtime_max_abs": 6.639331817659411e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.003960998142614711, "hypatia": 0.0037210133969521747, "sagan": 0.017751962083385725, "socrates": 0.0022569087104529497, "darwin": 0.003084828455212717, "einstein": 0.9692242892113817}, "confidence": 0.9016771056051921, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 173.20461200142745, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "56ba92fcf591d9e229f3f8ad5237c5869776100cf3a70c6bec64ce9064009b29", "compiled_ids_sha256": "6a4869b0ad66deb5f8da671c706b01b51b4f0ded31173e49dacf6f558a58b9e2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.125, 14.0625, 15.625, 13.5625, 13.875, 19.625], "runtime_probabilities": [0.0039609987288713455, 0.0037210138980299234, 0.017751963809132576, 0.002256908919662237, 0.003084828844293952, 0.9692243933677673], "decide_seconds": 0.1697679179997067, "probabilities_f64_minus_runtime_max_abs": 1.0415638562744078e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0036559613437314236, "darwin": 0.0019568950914558886, "sagan": 0.036923901561872675, "ibn_sina": 0.0019568950914558886, "einstein": 0.9522799723496722, "socrates": 0.0032263745618119964}, "confidence": 0.8706289823313822, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 168.98312400007853, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "c78dcb949364652ccb81e7ece0a7c58511b69ae2b059a4200cc368f7cd710b86", "compiled_ids_sha256": "d0d27afdbcce55ed017cb26871c2c98e4051b37964751abc8bd378765be94ee5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.6875, 16.625, 13.6875, 19.875, 14.1875], "runtime_probabilities": [0.003655961249023676, 0.0019568949937820435, 0.03692390024662018, 0.0019568949937820435, 0.9522799253463745, 0.0032263745088130236], "decide_seconds": 0.16547060100128874, "probabilities_f64_minus_runtime_max_abs": 4.7003297698466895e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004522467111022192, "einstein": 0.9765809704830842, "darwin": 0.001379274540924254, "sagan": 0.005806962716686877, "hypatia": 0.009574062949164255, "socrates": 0.002136262199118169}, "confidence": 0.9195301063414038, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 165.321311000298, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "8b89de27ed879d98c030e2f398520cd0febf51c68c5518567109e07953145424", "compiled_ids_sha256": "fbf019471853ab3512dfdfd5a3519d1703362c5a14b3eae89f8739cf6298fceb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 19.75, 13.1875, 14.625, 15.125, 13.625], "runtime_probabilities": [0.004522467497736216, 0.9765809774398804, 0.0013792746467515826, 0.005806962493807077, 0.009574064053595066, 0.0021362623665481806], "decide_seconds": 0.16186707800079603, "probabilities_f64_minus_runtime_max_abs": 6.9567961391570066e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.9081969960059988, "sagan": 0.07454934916738144, "socrates": 0.003950970761105045, "ibn_sina": 0.003077019122640834, "darwin": 0.004477036406058444, "hypatia": 0.0057486285368153825}, "confidence": 0.7909663741379316, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 170.94489900046028, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "c0960b215a98d11febcac1885a2697564706bea112c92b9221476c4435d5d8b9", "compiled_ids_sha256": "fc7df527b549d3694ff6cc0cf2bb2612d262bb2ac202c3ce0ad19182b48bc9ff", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 17.25, 14.3125, 14.0625, 14.4375, 14.6875], "runtime_probabilities": [0.9081971049308777, 0.0745493546128273, 0.003950971178710461, 0.0030770194716751575, 0.004477036651223898, 0.005748629104346037], "decide_seconds": 0.1675309450001805, "probabilities_f64_minus_runtime_max_abs": 1.0892487889435643e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0031127681511596557, "ibn_sina": 0.00452905093563427, "hypatia": 0.0020097541699079058, "einstein": 0.9780026807291319, "sagan": 0.010206373836512359, "darwin": 0.002139372177654028}, "confidence": 0.9237673777888181, "choice": "einstein"}}, "usage": {"input_tokens": 278, "output_tokens": 1}, "latency_ms": 169.03459400055, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 278, "compiled_prompt_sha256": "af19bfbc7bf8d777b190a6366cb8b8ef37d3d9885df8ab53705093e1856fea10", "compiled_ids_sha256": "5b52528a09013dbddda1f9e47550ee0c9ffc92cfb9f43beb8f1df7a3e7646fb1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 14.375, 13.5625, 19.75, 15.1875, 13.625], "runtime_probabilities": [0.0031127682887017727, 0.00452905148267746, 0.002009754069149494, 0.9780027270317078, 0.01020637433975935, 0.0021393722854554653], "decide_seconds": 0.16544037100175046, "probabilities_f64_minus_runtime_max_abs": 4.6302575884560326e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:31Z"} {"id": "c-hypatia-01-dessert-01", "uid": "c036", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "prompt_sha256_by_order": ["53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816"], "request_sha256": "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "compiled_prompt_sha256_by_order": ["4231600771fa3f6d01b11d9780088dab4c81fa8caa77568d942fc82a7a6b4f8c", "3e2f4297679effe8564f1df866abfb4406e82fc1c02552e3e8e74020d5222e3c", "0955081e6fd04107249a73b24edf436a83412f5c8545df9a1099bc8971404ccd", "845af03cd605ee29c93cdb5eb1a4d16ea4d5d00a134d31d4927d0ca50a95c13a", "820c72d1ae90bd487c7673c58e8ba149931905682bacacf34aef5a7b3f1e5485", "d7cbff19edacb7ae424f7fa26d318d9443bc71d83980e0de0903b52472c5bb5c"], "compiled_tokens_by_order": [395, 395, 395, 395, 395, 395], "choice": "socrates", "probs": {"darwin": 0.016691597680239964, "einstein": 0.012211833734953069, "hypatia": 0.261100518070963, "ibn_sina": 0.1397570362897121, "sagan": 0.1397570362897121, "socrates": 0.4304819779344199}, "confidence": 0.22667870117696431, "seconds": 0.18158258200067698, "seconds_all_orders": 1.0894954920040618, "seconds_by_order": [0.18341257300198777, 0.17948276300012367, 0.17692804300168063, 0.17681456400168827, 0.18960265499845264, 0.18325489400012884], "argmax_stable": false, "spread": {"darwin": 0.12944878868464355, "einstein": 0.10717299536875938, "hypatia": 0.39990347597872994, "ibn_sina": 0.18199807504319399, "sagan": 0.24720047791930946, "socrates": 0.31968833071182895}, "spread_2dp": {"darwin": 0.13, "einstein": 0.11, "hypatia": 0.4, "ibn_sina": 0.18, "sagan": 0.25, "socrates": 0.32}, "choices_by_order": ["socrates", "socrates", "socrates", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.016691597680239964, "einstein": 0.012211833734953069, "hypatia": 0.261100518070963, "ibn_sina": 0.1397570362897121, "sagan": 0.1397570362897121, "socrates": 0.4304819779344199}, {"ibn_sina": 0.11824106493453171, "hypatia": 0.15182453267216436, "sagan": 0.1720397343350569, "socrates": 0.3214125381891835, "darwin": 0.11824106493453171, "einstein": 0.11824106493453171}, {"hypatia": 0.1333859805737711, "darwin": 0.021774662752939003, "sagan": 0.0809026867938259, "ibn_sina": 0.24919781898508084, "einstein": 0.10388110612160013, "socrates": 0.410857744772783}, {"ibn_sina": 0.10501092025848725, "einstein": 0.011068069565772326, "darwin": 0.02655091863324038, "sagan": 0.038631358661591175, "hypatia": 0.533289456552501, "socrates": 0.2854492763284077}, {"einstein": 0.03063268910474062, "sagan": 0.018579665131470326, "socrates": 0.12896843830952576, "ibn_sina": 0.16559875273995608, "darwin": 0.14614038636488352, "hypatia": 0.5100800683494238}, {"socrates": 0.11079364722259094, "ibn_sina": 0.06719974394188687, "hypatia": 0.43819737517754354, "einstein": 0.031742911390907985, "sagan": 0.2657801430507798, "darwin": 0.08628617921629081}], "prompt_tokens": 395, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.016691597680239964, "einstein": 0.012211833734953069, "hypatia": 0.261100518070963, "ibn_sina": 0.1397570362897121, "sagan": 0.1397570362897121, "socrates": 0.4304819779344199}, "confidence": 0.22667870117696431, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 183.38457299978472, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "4231600771fa3f6d01b11d9780088dab4c81fa8caa77568d942fc82a7a6b4f8c", "compiled_ids_sha256": "7ef9a0540d1930f31f588689d49c705e105a918541d1c90201320219a03b21c8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 15.0625, 18.125, 17.5, 17.5, 18.625], "runtime_probabilities": [0.016691597178578377, 0.012211834080517292, 0.26110053062438965, 0.13975703716278076, 0.13975703716278076, 0.43048200011253357], "decide_seconds": 0.17879276900202967, "probabilities_f64_minus_runtime_max_abs": 2.217811367888345e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.11824106493453171, "hypatia": 0.15182453267216436, "sagan": 0.1720397343350569, "socrates": 0.3214125381891835, "darwin": 0.11824106493453171, "einstein": 0.11824106493453171}, "confidence": 0.04499109789739664, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 179.4577130021935, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "3e2f4297679effe8564f1df866abfb4406e82fc1c02552e3e8e74020d5222e3c", "compiled_ids_sha256": "1baafb93faf4b18a2b85ac937d9b7d0684e5f91e6451614c698e0f74463d5d5a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 17.25, 17.375, 18.0, 17.0, 17.0], "runtime_probabilities": [0.11824106425046921, 0.15182451903820038, 0.17203973233699799, 0.32141250371932983, 0.11824106425046921, 0.11824106425046921], "decide_seconds": 0.1749469179994776, "probabilities_f64_minus_runtime_max_abs": 3.446985369359723e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.1333859805737711, "darwin": 0.021774662752939003, "sagan": 0.0809026867938259, "ibn_sina": 0.24919781898508084, "einstein": 0.10388110612160013, "socrates": 0.410857744772783}, "confidence": 0.1614762659785579, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.90316299922415, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "0955081e6fd04107249a73b24edf436a83412f5c8545df9a1099bc8971404ccd", "compiled_ids_sha256": "2408a912179e8d8842eda565b9269ab95eb249ac1a73d511b205466a87a119fc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 15.6875, 17.0, 18.125, 17.25, 18.625], "runtime_probabilities": [0.13338598608970642, 0.021774664521217346, 0.08090268820524216, 0.2491978406906128, 0.1038811132311821, 0.41085776686668396], "decide_seconds": 0.17238113699932, "probabilities_f64_minus_runtime_max_abs": 2.209390098650843e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.10501092025848725, "einstein": 0.011068069565772326, "darwin": 0.02655091863324038, "sagan": 0.038631358661591175, "hypatia": 0.533289456552501, "socrates": 0.2854492763284077}, "confidence": 0.329324980563659, "choice": "hypatia"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.78991399952793, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "845af03cd605ee29c93cdb5eb1a4d16ea4d5d00a134d31d4927d0ca50a95c13a", "compiled_ids_sha256": "e435ee4c0012e1a2937895a9c9c205c811c1f490c1f4f0d18631555c828df116", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 14.875, 15.75, 16.125, 18.75, 18.125], "runtime_probabilities": [0.10501091182231903, 0.011068069376051426, 0.02655091881752014, 0.03863136097788811, 0.5332894325256348, 0.2854492962360382], "decide_seconds": 0.17232521800178802, "probabilities_f64_minus_runtime_max_abs": 2.4026866274340364e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03063268910474062, "sagan": 0.018579665131470326, "socrates": 0.12896843830952576, "ibn_sina": 0.16559875273995608, "darwin": 0.14614038636488352, "hypatia": 0.5100800683494238}, "confidence": 0.23695515295300662, "choice": "hypatia"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 189.57710500035319, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "820c72d1ae90bd487c7673c58e8ba149931905682bacacf34aef5a7b3f1e5485", "compiled_ids_sha256": "1add6aea2da9fb9d0a9c8833a85e5962baa4522d3edbf1acdc58ee030bbf50e2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.3125, 17.25, 17.5, 17.375, 18.625], "runtime_probabilities": [0.030632689595222473, 0.018579665571451187, 0.12896843254566193, 0.16559875011444092, 0.14614039659500122, 0.510080099105835], "decide_seconds": 0.18506094999975176, "probabilities_f64_minus_runtime_max_abs": 3.0756411195653754e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.11079364722259094, "ibn_sina": 0.06719974394188687, "hypatia": 0.43819737517754354, "einstein": 0.031742911390907985, "sagan": 0.2657801430507798, "darwin": 0.08628617921629081}, "confidence": 0.18523836101006985, "choice": "hypatia"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 183.22832400008338, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "d7cbff19edacb7ae424f7fa26d318d9443bc71d83980e0de0903b52472c5bb5c", "compiled_ids_sha256": "7eafaa910c92229237d1aa531c2512571a888d4741a40cfb06fa2e69ebbf824d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 16.5, 18.375, 15.75, 17.875, 16.75], "runtime_probabilities": [0.11079365015029907, 0.06719975173473358, 0.43819737434387207, 0.03174291178584099, 0.26578015089035034, 0.08628617227077484], "decide_seconds": 0.17829785199865, "probabilities_f64_minus_runtime_max_abs": 7.839570548284769e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:32Z"} {"id": "c-hypatia-01-dessert-02", "uid": "c037", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "prompt_sha256_by_order": ["60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b"], "request_sha256": "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "compiled_prompt_sha256_by_order": ["64cf4ddb66e4162fd4413e7c5eecfae78ed5607c7644b898fafd96f028bf8081", "126fd02f52a3255d24453db4c6d6615c2078c7e398d4dc621c4794f1dbc27888", "d4b5d824760b9b2b66b91bdb06cb7be6f8c7c4739faf8eeb587a17f85422db5d", "41469d816178843dac72662e56417911a5b5854d202c276b41a222ce9977785f", "48c60228de038271c453cf978f22cde2a0fb1bcf131fe7c1e2364daa12a7274c", "1fb528d3c58e26b2dbf098c2beb55874c4c20fc362e400b8bc58ba0599f2670d"], "compiled_tokens_by_order": [323, 323, 323, 323, 323, 323], "choice": "sagan", "probs": {"darwin": 0.024841363748491815, "einstein": 0.111331268403908, "hypatia": 0.05258916746823867, "ibn_sina": 0.008584945592879659, "sagan": 0.6406672150802412, "socrates": 0.16198603970624073}, "confidence": 0.37936112841064307, "seconds": 0.17352774366736412, "seconds_all_orders": 1.0411664620041847, "seconds_by_order": [0.17545636400245712, 0.17723698999907356, 0.17123163599899272, 0.1709029890007514, 0.17114630700234557, 0.17519217600056436], "argmax_stable": true, "spread": {"darwin": 0.03745086652566461, "einstein": 0.32118943832096675, "hypatia": 0.06827385741244749, "ibn_sina": 0.022022762013891625, "sagan": 0.5179567275797404, "socrates": 0.32940729013060804}, "spread_2dp": {"darwin": 0.030000000000000002, "einstein": 0.31999999999999995, "hypatia": 0.06, "ibn_sina": 0.019999999999999997, "sagan": 0.52, "socrates": 0.32999999999999996}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.024841363748491815, "einstein": 0.111331268403908, "hypatia": 0.05258916746823867, "ibn_sina": 0.008584945592879659, "sagan": 0.6406672150802412, "socrates": 0.16198603970624073}, {"ibn_sina": 0.02898255357460117, "hypatia": 0.016513761243651566, "sagan": 0.40009181092224594, "socrates": 0.1471855518193462, "darwin": 0.05414653855181145, "einstein": 0.35307978388834366}, {"hypatia": 0.025900865647637637, "darwin": 0.01780138727102674, "sagan": 0.35755042131617054, "ibn_sina": 0.024331611527567618, "einstein": 0.21686529292142695, "socrates": 0.35755042131617054}, {"ibn_sina": 0.0256572066066822, "einstein": 0.10147622062525281, "darwin": 0.03506924345454822, "sagan": 0.5153387706443698, "hypatia": 0.07902976008607353, "socrates": 0.2434287985830734}, {"einstein": 0.0582736212765543, "sagan": 0.7099180392699441, "socrates": 0.12336525721054617, "ibn_sina": 0.0069597915607095455, "darwin": 0.01669567202614684, "hypatia": 0.08478761865609905}, {"socrates": 0.0281431311855625, "ibn_sina": 0.008063142083329266, "hypatia": 0.02995820720362969, "einstein": 0.03189034556737691, "sagan": 0.875507148895911, "darwin": 0.026438025064190716}], "prompt_tokens": 323, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.024841363748491815, "einstein": 0.111331268403908, "hypatia": 0.05258916746823867, "ibn_sina": 0.008584945592879659, "sagan": 0.6406672150802412, "socrates": 0.16198603970624073}, "confidence": 0.37936112841064307, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 175.42796400084626, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "64cf4ddb66e4162fd4413e7c5eecfae78ed5607c7644b898fafd96f028bf8081", "compiled_ids_sha256": "2edffcb25f65010db4bf8ec46a80e7f11423c9168ec99e3b3ac49296338f2c48", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 17.125, 16.375, 14.5625, 18.875, 17.5], "runtime_probabilities": [0.024841362610459328, 0.11133125424385071, 0.05258916690945625, 0.0085849454626441, 0.640667200088501, 0.16198603808879852], "decide_seconds": 0.17137785500017344, "probabilities_f64_minus_runtime_max_abs": 1.4991740182601632e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02898255357460117, "hypatia": 0.016513761243651566, "sagan": 0.40009181092224594, "socrates": 0.1471855518193462, "darwin": 0.05414653855181145, "einstein": 0.35307978388834366}, "confidence": 0.24968073393713563, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 177.21295100272982, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "126fd02f52a3255d24453db4c6d6615c2078c7e398d4dc621c4794f1dbc27888", "compiled_ids_sha256": "a9dcef5a25e17758145e02113f8a18e8cf23c6d68f0a239da3291718a3d2435b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 15.1875, 18.375, 17.375, 16.375, 18.25], "runtime_probabilities": [0.028982553631067276, 0.01651376113295555, 0.4000917971134186, 0.1471855640411377, 0.05414653569459915, 0.35307976603507996], "decide_seconds": 0.17332565099786734, "probabilities_f64_minus_runtime_max_abs": 1.785326370074003e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.025900865647637637, "darwin": 0.01780138727102674, "sagan": 0.35755042131617054, "ibn_sina": 0.024331611527567618, "einstein": 0.21686529292142695, "socrates": 0.35755042131617054}, "confidence": 0.261230819505071, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 171.20848600097816, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "d4b5d824760b9b2b66b91bdb06cb7be6f8c7c4739faf8eeb587a17f85422db5d", "compiled_ids_sha256": "55e8c17d426b8b4a5a6226037d4a13fa22099e749ae06aab91c690a6d167adaa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 15.375, 18.375, 15.6875, 17.875, 18.375], "runtime_probabilities": [0.025900866836309433, 0.01780138723552227, 0.3575504422187805, 0.02433161251246929, 0.21686530113220215, 0.3575504422187805], "decide_seconds": 0.16732325699922512, "probabilities_f64_minus_runtime_max_abs": 2.0902609976225506e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0256572066066822, "einstein": 0.10147622062525281, "darwin": 0.03506924345454822, "sagan": 0.5153387706443698, "hypatia": 0.07902976008607353, "socrates": 0.2434287985830734}, "confidence": 0.25782322931214097, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.87947899926803, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "41469d816178843dac72662e56417911a5b5854d202c276b41a222ce9977785f", "compiled_ids_sha256": "ad858089a65d8558ae08ba6a32b72aa795e8595e040ade53729b92e1dfa8ee88", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 17.0, 15.9375, 18.625, 16.75, 17.875], "runtime_probabilities": [0.025657204911112785, 0.10147621482610703, 0.035069242119789124, 0.5153387784957886, 0.07902976125478745, 0.2434287965297699], "decide_seconds": 0.16703650900308276, "probabilities_f64_minus_runtime_max_abs": 7.851418737381266e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0582736212765543, "sagan": 0.7099180392699441, "socrates": 0.12336525721054617, "ibn_sina": 0.0069597915607095455, "darwin": 0.01669567202614684, "hypatia": 0.08478761865609905}, "confidence": 0.45352569968534195, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 171.12250699938159, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "48c60228de038271c453cf978f22cde2a0fb1bcf131fe7c1e2364daa12a7274c", "compiled_ids_sha256": "aad3d2601ba3acda91bd63406dc2e8d9f7609b44ed309b8847048d312197eeaa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 18.875, 17.125, 14.25, 15.125, 16.75], "runtime_probabilities": [0.05827362462878227, 0.7099180817604065, 0.12336525321006775, 0.006959792226552963, 0.016695674508810043, 0.08478762954473495], "decide_seconds": 0.16719011799796135, "probabilities_f64_minus_runtime_max_abs": 4.249046237081444e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0281431311855625, "ibn_sina": 0.008063142083329266, "hypatia": 0.02995820720362969, "einstein": 0.03189034556737691, "sagan": 0.875507148895911, "darwin": 0.026438025064190716}, "confidence": 0.6836803319767997, "choice": "sagan"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 175.16877600064618, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "1fb528d3c58e26b2dbf098c2beb55874c4c20fc362e400b8bc58ba0599f2670d", "compiled_ids_sha256": "2e7101de9cfe6c101b9944ff7428dc9155907f17e6fd68c36d0d043c55b402cc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 14.4375, 15.75, 15.8125, 19.125, 15.625], "runtime_probabilities": [0.028143135830760002, 0.008063143119215965, 0.02995821088552475, 0.031890351325273514, 0.8755072355270386, 0.0264380294829607], "decide_seconds": 0.17122251600085292, "probabilities_f64_minus_runtime_max_abs": 8.663112760487479e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:33Z"} {"id": "c-hypatia-01-dessert-02", "uid": "c038", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "prompt_sha256_by_order": ["cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70"], "request_sha256": "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "compiled_prompt_sha256_by_order": ["b813667369f4e70abaa4bbd36b3b8b690e78c38b9ec85f4f3d5388a491e9ce79", "2f940aa121ab8e63c8413de9a7abfaaf942772de826f2b4bbae41625a4667b75", "4b7a7ef91594b7dff62c49741edbfe626704376c8c9f2323971eb2a90b1660eb", "69047759db4cdc0a6aac6d5730c13eb5159344815b3f54af7c431b13ededc1d5", "fee024091133f61a5a8e7910bb908002435ca23f94302179c6ce66d24dc46941", "ee0f15ee9be62c89613213d7c0e9b66d4048e351ebdb9dfc8b68082c09a24ed8"], "compiled_tokens_by_order": [274, 274, 274, 274, 274, 274], "choice": "sagan", "probs": {"darwin": 0.057322717154951554, "einstein": 0.017482440987958058, "hypatia": 0.20007594212436225, "ibn_sina": 0.025436801544048322, "sagan": 0.5438627977884775, "socrates": 0.1558193004002024}, "confidence": 0.29070805955909473, "seconds": 0.1666795165001531, "seconds_all_orders": 1.0000770990009187, "seconds_by_order": [0.16582126800130936, 0.1654371210024692, 0.16455146799853537, 0.16808279099859647, 0.16683998000007705, 0.16934447099993122], "argmax_stable": false, "spread": {"darwin": 0.17953301479767028, "einstein": 0.16727347414706906, "hypatia": 0.2733006772434574, "ibn_sina": 0.020143176253146426, "sagan": 0.8158427603199487, "socrates": 0.4210033531423523}, "spread_2dp": {"darwin": 0.18, "einstein": 0.16999999999999998, "hypatia": 0.27, "ibn_sina": 0.019999999999999997, "sagan": 0.8200000000000001, "socrates": 0.42000000000000004}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.057322717154951554, "einstein": 0.017482440987958058, "hypatia": 0.20007594212436225, "ibn_sina": 0.025436801544048322, "sagan": 0.5438627977884775, "socrates": 0.1558193004002024}, {"ibn_sina": 0.01796794933999097, "hypatia": 0.0066100391621926645, "sagan": 0.7177272186418064, "socrates": 0.040491398892554264, "darwin": 0.035733534103996455, "einstein": 0.18146985985945924}, {"hypatia": 0.023405883091081065, "darwin": 0.00713839102924527, "sagan": 0.8782992832988031, "ibn_sina": 0.013336270182957919, "einstein": 0.01419638571239017, "socrates": 0.06362378668552249}, {"ibn_sina": 0.030036398321397827, "einstein": 0.015103249564512191, "darwin": 0.06768806849986243, "sagan": 0.5667452587186843, "hypatia": 0.05271552075228376, "socrates": 0.26771150414325945}, {"einstein": 0.03788179608572182, "sagan": 0.06245652297885443, "socrates": 0.4614947520349066, "ibn_sina": 0.026035752311062425, "darwin": 0.1322204601838047, "hypatia": 0.2799107164056501}, {"socrates": 0.06867257245829321, "ibn_sina": 0.009893222068251403, "hypatia": 0.14537983703504184, "einstein": 0.014394553172217118, "sagan": 0.5749884094392809, "darwin": 0.18667140582691555}], "prompt_tokens": 274, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.057322717154951554, "einstein": 0.017482440987958058, "hypatia": 0.20007594212436225, "ibn_sina": 0.025436801544048322, "sagan": 0.5438627977884775, "socrates": 0.1558193004002024}, "confidence": 0.29070805955909473, "choice": "sagan"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 165.79615799855674, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "b813667369f4e70abaa4bbd36b3b8b690e78c38b9ec85f4f3d5388a491e9ce79", "compiled_ids_sha256": "1356d08876a1c4cf2b6410ee614660ec7bf937bb468b506992fcb7bf08d22f64", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 15.4375, 17.875, 15.8125, 18.875, 17.625], "runtime_probabilities": [0.05732271447777748, 0.017482440918684006, 0.20007595419883728, 0.025436801835894585, 0.5438627600669861, 0.15581929683685303], "decide_seconds": 0.16234115499901236, "probabilities_f64_minus_runtime_max_abs": 3.7721491374931304e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01796794933999097, "hypatia": 0.0066100391621926645, "sagan": 0.7177272186418064, "socrates": 0.040491398892554264, "darwin": 0.035733534103996455, "einstein": 0.18146985985945924}, "confidence": 0.496561051224805, "choice": "sagan"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 165.41480099840555, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "2f940aa121ab8e63c8413de9a7abfaaf942772de826f2b4bbae41625a4667b75", "compiled_ids_sha256": "e3628eef31f404db4947d62516ea71b5897d6d12c87f09aa96e9df8effe4ea81", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.4375, 19.125, 16.25, 16.125, 17.75], "runtime_probabilities": [0.0179679486900568, 0.006610038690268993, 0.7177271842956543, 0.04049139842391014, 0.035733532160520554, 0.1814698576927185], "decide_seconds": 0.16201117799937492, "probabilities_f64_minus_runtime_max_abs": 3.4346152144593134e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.023405883091081065, "darwin": 0.00713839102924527, "sagan": 0.8782992832988031, "ibn_sina": 0.013336270182957919, "einstein": 0.01419638571239017, "socrates": 0.06362378668552249}, "confidence": 0.7039860099999253, "choice": "sagan"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 164.53066800022498, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "4b7a7ef91594b7dff62c49741edbfe626704376c8c9f2323971eb2a90b1660eb", "compiled_ids_sha256": "ea6455a950ef6d665f631c4a2ef5673566a510a7f3cc337c4d782173ebd523c2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 14.9375, 19.75, 15.5625, 15.625, 17.125], "runtime_probabilities": [0.023405881598591805, 0.007138391491025686, 0.8782992362976074, 0.013336269184947014, 0.014196384698152542, 0.06362378597259521], "decide_seconds": 0.1612141330006125, "probabilities_f64_minus_runtime_max_abs": 4.700119571321437e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.030036398321397827, "einstein": 0.015103249564512191, "darwin": 0.06768806849986243, "sagan": 0.5667452587186843, "hypatia": 0.05271552075228376, "socrates": 0.26771150414325945}, "confidence": 0.34106853579419805, "choice": "sagan"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 168.06208100024378, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "69047759db4cdc0a6aac6d5730c13eb5159344815b3f54af7c431b13ededc1d5", "compiled_ids_sha256": "bc26a3e9a780ff4dad9df25142277a157aedbffcc82d966cf5e573f8cc815d17", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.125, 16.625, 18.75, 16.375, 18.0], "runtime_probabilities": [0.03003639727830887, 0.015103247947990894, 0.06768806278705597, 0.5667452216148376, 0.05271551385521889, 0.26771149039268494], "decide_seconds": 0.16473685699747875, "probabilities_f64_minus_runtime_max_abs": 3.7103846661779016e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03788179608572182, "sagan": 0.06245652297885443, "socrates": 0.4614947520349066, "ibn_sina": 0.026035752311062425, "darwin": 0.1322204601838047, "hypatia": 0.2799107164056501}, "confidence": 0.2337220046991162, "choice": "socrates"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 166.8174699989322, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "fee024091133f61a5a8e7910bb908002435ca23f94302179c6ce66d24dc46941", "compiled_ids_sha256": "98c41af6d43a5b89eef460919ec169e6f17c46828190e5d394cace8da14cbb84", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 16.5, 18.5, 15.625, 17.25, 18.0], "runtime_probabilities": [0.037881795316934586, 0.06245651841163635, 0.46149471402168274, 0.026035752147436142, 0.13222046196460724, 0.27991071343421936], "decide_seconds": 0.16336330600097426, "probabilities_f64_minus_runtime_max_abs": 3.801322384644479e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.06867257245829321, "ibn_sina": 0.009893222068251403, "hypatia": 0.14537983703504184, "einstein": 0.014394553172217118, "sagan": 0.5749884094392809, "darwin": 0.18667140582691555}, "confidence": 0.32886731173236683, "choice": "sagan"}}, "usage": {"input_tokens": 274, "output_tokens": 1}, "latency_ms": 169.32207100035157, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 274, "compiled_prompt_sha256": "ee0f15ee9be62c89613213d7c0e9b66d4048e351ebdb9dfc8b68082c09a24ed8", "compiled_ids_sha256": "9e6f6d29157bceafa71e5e20a62b05c99a10d78c1f78f4112053f0f3eee46ce7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 14.8125, 17.5, 15.1875, 18.875, 17.75], "runtime_probabilities": [0.06867257505655289, 0.009893221780657768, 0.14537984132766724, 0.014394554309546947, 0.5749884247779846, 0.1866714060306549], "decide_seconds": 0.1657850679985131, "probabilities_f64_minus_runtime_max_abs": 1.5338703751588412e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:34Z"} {"id": "c-hypatia-01-dessert-03", "uid": "c039", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "prompt_sha256_by_order": ["a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e"], "request_sha256": "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "compiled_prompt_sha256_by_order": ["21829cbfc35a4be0d7b256334b6e53727cf63ff22411b3b96a97b55be3b983e1", "d14223f8e924618fd2fa5162ab6bfe1f68c2a20982377dc8992042d6356ac0bc", "90524ea2154ee3db8ecd561b5c8cd5cb035d53cb94f0bb90fe604bbbd1b75f2f", "a252ebbab08bae3d0a4a48241c07c96c8f8536db39c9c0d31d670d252e85d551", "7601ee6d28c81e741632eab0d2d4f312a74d4ad024c7beb8c48c484ba9059fc0", "0a8c432afe9dacff969a9d99c9a7d6e411357522787a2c272bee56910b8c2d8e"], "compiled_tokens_by_order": [298, 298, 298, 298, 298, 298], "choice": "socrates", "probs": {"darwin": 0.018724894329085997, "einstein": 0.027244560488453168, "hypatia": 0.21429506764252984, "ibn_sina": 0.027244560488453168, "sagan": 0.129976528750387, "socrates": 0.5825143883010908}, "confidence": 0.34092659690189686, "seconds": 0.16739874416634848, "seconds_all_orders": 1.0043924649980909, "seconds_by_order": [0.17231111799992505, 0.1658465079999587, 0.1651375029978226, 0.16516108300129417, 0.16518115299913916, 0.1707550999999512], "argmax_stable": false, "spread": {"darwin": 0.07963720833924293, "einstein": 0.18128939931546773, "hypatia": 0.549698586100251, "ibn_sina": 0.024261170711447025, "sagan": 0.6832690226314274, "socrates": 0.6254255662108399}, "spread_2dp": {"darwin": 0.08, "einstein": 0.18000000000000002, "hypatia": 0.5499999999999999, "ibn_sina": 0.019999999999999997, "sagan": 0.6799999999999999, "socrates": 0.63}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "hypatia", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.018724894329085997, "einstein": 0.027244560488453168, "hypatia": 0.21429506764252984, "ibn_sina": 0.027244560488453168, "sagan": 0.129976528750387, "socrates": 0.5825143883010908}, {"ibn_sina": 0.033060317899612654, "hypatia": 0.033060317899612654, "sagan": 0.12283398253668332, "socrates": 0.5505037170007504, "darwin": 0.0580226648903031, "einstein": 0.2025189997730379}, {"hypatia": 0.03752225907201763, "darwin": 0.017724260168904233, "sagan": 0.13096555270007543, "ibn_sina": 0.03311327740903229, "einstein": 0.11557669460309616, "socrates": 0.6650979560468742}, {"ibn_sina": 0.019943363988152368, "einstein": 0.021229600457570177, "darwin": 0.01553190749105112, "sagan": 0.35350458509868543, "hypatia": 0.1892173692079353, "socrates": 0.4005731737566056}, {"einstein": 0.034997392442745454, "sagan": 0.01994091312424416, "socrates": 0.27527581456154754, "ibn_sina": 0.025604639283491666, "darwin": 0.06142233658810757, "hypatia": 0.5827589039998636}, {"socrates": 0.03967238983603428, "ibn_sina": 0.008852106697585268, "hypatia": 0.12219956360979724, "einstein": 0.03089688827061754, "sagan": 0.7032099357556716, "darwin": 0.09516911583029404}], "prompt_tokens": 298, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.018724894329085997, "einstein": 0.027244560488453168, "hypatia": 0.21429506764252984, "ibn_sina": 0.027244560488453168, "sagan": 0.129976528750387, "socrates": 0.5825143883010908}, "confidence": 0.34092659690189686, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 172.28429799797595, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "21829cbfc35a4be0d7b256334b6e53727cf63ff22411b3b96a97b55be3b983e1", "compiled_ids_sha256": "90ee18b88d0e89b3999ace8ac624b7a7824208a64fcb7a98548c6bbcf7208086", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 15.5625, 17.625, 15.5625, 17.125, 18.625], "runtime_probabilities": [0.018724896013736725, 0.027244562283158302, 0.21429508924484253, 0.027244562283158302, 0.1299765259027481, 0.5825144052505493], "decide_seconds": 0.16855670700169867, "probabilities_f64_minus_runtime_max_abs": 2.1602312688662195e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.033060317899612654, "hypatia": 0.033060317899612654, "sagan": 0.12283398253668332, "socrates": 0.5505037170007504, "darwin": 0.0580226648903031, "einstein": 0.2025189997730379}, "confidence": 0.2743407522640454, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.82331799872918, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "d14223f8e924618fd2fa5162ab6bfe1f68c2a20982377dc8992042d6356ac0bc", "compiled_ids_sha256": "3114c0291676974a6225f6099745086841a066740ff190221b427db42ef5d97f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.5625, 16.875, 18.375, 16.125, 17.375], "runtime_probabilities": [0.03306031599640846, 0.03306031599640846, 0.1228339821100235, 0.5505037307739258, 0.05802266299724579, 0.2025190144777298], "decide_seconds": 0.1619489370023075, "probabilities_f64_minus_runtime_max_abs": 1.4704691908562495e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.03752225907201763, "darwin": 0.017724260168904233, "sagan": 0.13096555270007543, "ibn_sina": 0.03311327740903229, "einstein": 0.11557669460309616, "socrates": 0.6650979560468742}, "confidence": 0.38922273078091707, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.1156930020079, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "90524ea2154ee3db8ecd561b5c8cd5cb035d53cb94f0bb90fe604bbbd1b75f2f", "compiled_ids_sha256": "0420faee76bd77bf67040c2f86f5e8a070532b9429aff16201fa2a787f279c4c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 15.25, 17.25, 15.875, 17.125, 18.875], "runtime_probabilities": [0.03752226382493973, 0.017724260687828064, 0.13096554577350616, 0.033113278448581696, 0.11557668447494507, 0.6650979518890381], "decide_seconds": 0.16151623199766618, "probabilities_f64_minus_runtime_max_abs": 1.012815109557863e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.019943363988152368, "einstein": 0.021229600457570177, "darwin": 0.01553190749105112, "sagan": 0.35350458509868543, "hypatia": 0.1892173692079353, "socrates": 0.4005731737566056}, "confidence": 0.28917319108690265, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.13924299943028, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "a252ebbab08bae3d0a4a48241c07c96c8f8536db39c9c0d31d670d252e85d551", "compiled_ids_sha256": "a822ae5c78c7b9e78d08a258d35989a853a092f021918f67a037a062492b64eb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 15.3125, 15.0, 18.125, 17.5, 18.25], "runtime_probabilities": [0.019943363964557648, 0.02122960053384304, 0.01553190778940916, 0.35350459814071655, 0.18921737372875214, 0.40057316422462463], "decide_seconds": 0.16152080100073363, "probabilities_f64_minus_runtime_max_abs": 1.3042031121734965e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.034997392442745454, "sagan": 0.01994091312424416, "socrates": 0.27527581456154754, "ibn_sina": 0.025604639283491666, "darwin": 0.06142233658810757, "hypatia": 0.5827589039998636}, "confidence": 0.36912062954754954, "choice": "hypatia"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.15908400106127, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "7601ee6d28c81e741632eab0d2d4f312a74d4ad024c7beb8c48c484ba9059fc0", "compiled_ids_sha256": "6b2bf7c2998bfb19b334c181fff999c6c1610dd21800c566bb5f90f23076b154", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.25, 17.875, 15.5, 16.375, 18.625], "runtime_probabilities": [0.03499739244580269, 0.01994091272354126, 0.2752758264541626, 0.02560463920235634, 0.061422333121299744, 0.582758903503418], "decide_seconds": 0.16150690100039355, "probabilities_f64_minus_runtime_max_abs": 1.1892615059405642e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.03967238983603428, "ibn_sina": 0.008852106697585268, "hypatia": 0.12219956360979724, "einstein": 0.03089688827061754, "sagan": 0.7032099357556716, "darwin": 0.09516911583029404}, "confidence": 0.4387493678474549, "choice": "sagan"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 170.73304099903908, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "0a8c432afe9dacff969a9d99c9a7d6e411357522787a2c272bee56910b8c2d8e", "compiled_ids_sha256": "df188a4504ecc2073803b7e34e13219aa8bf8c279cf1ab309a088df460523472", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 14.5, 17.125, 15.75, 18.875, 16.875], "runtime_probabilities": [0.03967239335179329, 0.00885210745036602, 0.12219955772161484, 0.03089689090847969, 0.7032099962234497, 0.09516911953687668], "decide_seconds": 0.1671520279996912, "probabilities_f64_minus_runtime_max_abs": 6.046777811796744e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:35Z"} {"id": "c-hypatia-01-main-01", "uid": "c040", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "prompt_sha256_by_order": ["6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372"], "request_sha256": "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "compiled_prompt_sha256_by_order": ["a85f301fbcbe5359d5137e8ce9e30a12f45037248d0741358fb7ff25b1b3eafd", "b61c5ac999929626f974063e57c72670796723e26bac9fb12e40f33023fb5628", "081929d17758424a25168221bbd1848ce541fe1930acd1433380e80eb2c816a5", "02542a5ccc36f918849687042233556561fb2d6c471f687f7a179dbb318c5df6", "fbab03d13c25e0f115e70de830408c8fb2989d1e6e51aee8b7ea316b12426d1d", "ceffa6eefc3da573362e5e4396ae6d455b2ab4700f3b030698f74aa75fc83d91"], "compiled_tokens_by_order": [368, 368, 368, 368, 368, 368], "choice": "hypatia", "probs": {"darwin": 0.028870025727721713, "einstein": 0.018639889569038854, "hypatia": 0.39853841191200007, "ibn_sina": 0.31038202728109976, "sagan": 0.11418316674583034, "socrates": 0.12938647876430914}, "confidence": 0.20820615059223835, "seconds": 0.1742258768335887, "seconds_all_orders": 1.0453552610015322, "seconds_by_order": [0.17185375199915143, 0.17320804200062412, 0.1710497669992037, 0.17626536800162285, 0.17491245799828903, 0.17806587400264107], "argmax_stable": false, "spread": {"darwin": 0.029072335006820663, "einstein": 0.09113040820490828, "hypatia": 0.4715795265566062, "ibn_sina": 0.4281024547165794, "sagan": 0.10705686626306687, "socrates": 0.3067562117243971}, "spread_2dp": {"darwin": 0.030000000000000002, "einstein": 0.09, "hypatia": 0.47000000000000003, "ibn_sina": 0.43, "sagan": 0.1, "socrates": 0.31}, "choices_by_order": ["hypatia", "hypatia", "ibn_sina", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.028870025727721713, "einstein": 0.018639889569038854, "hypatia": 0.39853841191200007, "ibn_sina": 0.31038202728109976, "sagan": 0.11418316674583034, "socrates": 0.12938647876430914}, {"ibn_sina": 0.06385388906467433, "hypatia": 0.4163796607203662, "sagan": 0.03872934145961423, "socrates": 0.3674537608849479, "darwin": 0.04972945880572298, "einstein": 0.06385388906467433}, {"hypatia": 0.18098062483099356, "darwin": 0.024493064121841653, "sagan": 0.05185181715283409, "ibn_sina": 0.4919563437812537, "einstein": 0.10977029777394713, "socrates": 0.14094785233912993}, {"ibn_sina": 0.0817360609333667, "einstein": 0.01941401614252953, "darwin": 0.049575426960226904, "sagan": 0.09261909100642032, "hypatia": 0.6039523395422604, "socrates": 0.15270306541519613}, {"einstein": 0.026477418387473604, "sagan": 0.007126300482763476, "socrates": 0.11866355659790856, "ibn_sina": 0.19564312981990037, "darwin": 0.049466329865639154, "hypatia": 0.6026232648463148}, {"socrates": 0.06069754916055083, "ibn_sina": 0.11339795084524268, "hypatia": 0.6525601513875997, "einstein": 0.019705609097577556, "sagan": 0.10007334038036686, "darwin": 0.05356539912866232}], "prompt_tokens": 368, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.028870025727721713, "einstein": 0.018639889569038854, "hypatia": 0.39853841191200007, "ibn_sina": 0.31038202728109976, "sagan": 0.11418316674583034, "socrates": 0.12938647876430914}, "confidence": 0.20820615059223835, "choice": "hypatia"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 171.82515199965565, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "a85f301fbcbe5359d5137e8ce9e30a12f45037248d0741358fb7ff25b1b3eafd", "compiled_ids_sha256": "c6dbc71f4fcb0e3fc82bd7695103d89d45b3a9f018a7f14f1ed0ba8bed8dbf04", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 15.9375, 19.0, 18.75, 17.75, 17.875], "runtime_probabilities": [0.028870027512311935, 0.01863989047706127, 0.39853841066360474, 0.31038203835487366, 0.11418317258358002, 0.12938648462295532], "decide_seconds": 0.16731249599979492, "probabilities_f64_minus_runtime_max_abs": 1.1073773897862793e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.06385388906467433, "hypatia": 0.4163796607203662, "sagan": 0.03872934145961423, "socrates": 0.3674537608849479, "darwin": 0.04972945880572298, "einstein": 0.06385388906467433}, "confidence": 0.24141666203441703, "choice": "hypatia"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 173.18305200024042, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "b61c5ac999929626f974063e57c72670796723e26bac9fb12e40f33023fb5628", "compiled_ids_sha256": "8cedba15c25a3ae8f2644bdcb4b33b7308817bd310b112521064579da125c625", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 18.875, 16.5, 18.75, 16.75, 17.0], "runtime_probabilities": [0.06385388970375061, 0.4163796603679657, 0.038729339838027954, 0.36745375394821167, 0.04972946271300316, 0.06385388970375061], "decide_seconds": 0.16884710500016809, "probabilities_f64_minus_runtime_max_abs": 6.936736240970021e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.18098062483099356, "darwin": 0.024493064121841653, "sagan": 0.05185181715283409, "ibn_sina": 0.4919563437812537, "einstein": 0.10977029777394713, "socrates": 0.14094785233912993}, "confidence": 0.20673952566509135, "choice": "ibn_sina"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 171.02538800099865, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "081929d17758424a25168221bbd1848ce541fe1930acd1433380e80eb2c816a5", "compiled_ids_sha256": "dcc783e8b58640f809027b45c0b3df1e3a26c539c0d3ac436a4aa6ec9f9a1260", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.125, 16.125, 16.875, 19.125, 17.625, 17.875], "runtime_probabilities": [0.1809806376695633, 0.024493062868714333, 0.05185181275010109, 0.49195632338523865, 0.10977029800415039, 0.1409478485584259], "decide_seconds": 0.166766390997509, "probabilities_f64_minus_runtime_max_abs": 2.0396015043555593e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0817360609333667, "einstein": 0.01941401614252953, "darwin": 0.049575426960226904, "sagan": 0.09261909100642032, "hypatia": 0.6039523395422604, "socrates": 0.15270306541519613}, "confidence": 0.3068076971968644, "choice": "hypatia"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 176.24108800009708, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "02542a5ccc36f918849687042233556561fb2d6c471f687f7a179dbb318c5df6", "compiled_ids_sha256": "5d8fe48e10ed3cd7d04f6fe7c9bc2336ad464c766699d896f7a2b8e368b4a1a2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 15.6875, 16.625, 17.25, 19.125, 17.75], "runtime_probabilities": [0.08173605799674988, 0.019414016976952553, 0.04957542568445206, 0.09261909127235413, 0.6039523482322693, 0.15270306169986725], "decide_seconds": 0.17198884100071155, "probabilities_f64_minus_runtime_max_abs": 8.690008934664206e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.026477418387473604, "sagan": 0.007126300482763476, "socrates": 0.11866355659790856, "ibn_sina": 0.19564312981990037, "darwin": 0.049466329865639154, "hypatia": 0.6026232648463148}, "confidence": 0.3540312851695926, "choice": "hypatia"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 174.8875279990898, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "fbab03d13c25e0f115e70de830408c8fb2989d1e6e51aee8b7ea316b12426d1d", "compiled_ids_sha256": "d0a7b7e25dce6d1359fbebcd1b4ae2af4c2c6e0ff3365d78e4e438a0f0168f70", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 14.9375, 17.75, 18.25, 16.875, 19.375], "runtime_probabilities": [0.026477420702576637, 0.007126300595700741, 0.11866355687379837, 0.19564314186573029, 0.0494663342833519, 0.6026232838630676], "decide_seconds": 0.17053478200250538, "probabilities_f64_minus_runtime_max_abs": 1.9016752839107198e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.06069754916055083, "ibn_sina": 0.11339795084524268, "hypatia": 0.6525601513875997, "einstein": 0.019705609097577556, "sagan": 0.10007334038036686, "darwin": 0.05356539912866232}, "confidence": 0.35260523857528403, "choice": "hypatia"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 178.04193399933865, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "ceffa6eefc3da573362e5e4396ae6d455b2ab4700f3b030698f74aa75fc83d91", "compiled_ids_sha256": "38f951a9aa964795d4df957a23f7e6e3286f5fb42a4d79c7acee8884325aaa48", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 17.5, 19.25, 15.75, 17.375, 16.75], "runtime_probabilities": [0.06069754809141159, 0.11339794099330902, 0.6525601744651794, 0.01970561034977436, 0.10007334500551224, 0.05356540158390999], "decide_seconds": 0.1738184360001469, "probabilities_f64_minus_runtime_max_abs": 2.3077579736074938e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:36Z"} {"id": "c-hypatia-01-main-01", "uid": "c041", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "prompt_sha256_by_order": ["c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c"], "request_sha256": "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "compiled_prompt_sha256_by_order": ["937427227f5f99118cb41f68bee040c493d9bec3da24346374497e87fd4b0c16", "17b90d81647891d8db0520a988ad0cdae1e9f5bd1fed5dcd58e150aedbf9bc03", "0f3022a27181d5a0a95921e9f56219e94b652cc915de14ff442d6bba8f27fe29", "58f0b52eb9761ac361ed3c74d0385ea3ec690a657f7e3ee4907d149256845dab", "3a18fbc6a84be5fcb500032684a23491003db5acc02f325e46b5f30aead3dee2", "dd944680d9992422a25d628a6e81a6b820f291626cc233ff3d8ba7afc2cc8b85"], "compiled_tokens_by_order": [297, 297, 297, 297, 297, 297], "choice": "ibn_sina", "probs": {"darwin": 0.09905159933815753, "einstein": 0.013405176251469027, "hypatia": 0.20969223744439147, "ibn_sina": 0.5030255277420331, "sagan": 0.08741272961197438, "socrates": 0.08741272961197438}, "confidence": 0.22640719514043184, "seconds": 0.17411764449934708, "seconds_all_orders": 1.0447058669960825, "seconds_by_order": [0.16755276500043692, 0.16557208999802242, 0.21409910800139187, 0.16558238899960998, 0.16642169399710838, 0.16547782099951291], "argmax_stable": false, "spread": {"darwin": 0.42775481127962867, "einstein": 0.30496129231735497, "hypatia": 0.3486721297333572, "ibn_sina": 0.6072578217156172, "sagan": 0.2383898920471388, "socrates": 0.1369776787181444}, "spread_2dp": {"darwin": 0.43000000000000005, "einstein": 0.31, "hypatia": 0.35, "ibn_sina": 0.6, "sagan": 0.22999999999999998, "socrates": 0.13999999999999999}, "choices_by_order": ["ibn_sina", "einstein", "ibn_sina", "hypatia", "hypatia", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.09905159933815753, "einstein": 0.013405176251469027, "hypatia": 0.20969223744439147, "ibn_sina": 0.5030255277420331, "sagan": 0.08741272961197438, "socrates": 0.08741272961197438}, {"ibn_sina": 0.11712047854482453, "hypatia": 0.13271488908551432, "sagan": 0.08049564923072598, "socrates": 0.10335845954503321, "darwin": 0.24794405502507794, "einstein": 0.318366468568824}, {"hypatia": 0.12587807378618757, "darwin": 0.028087194763087894, "sagan": 0.013267451366405308, "ibn_sina": 0.7243783002604417, "einstein": 0.021874329275771914, "socrates": 0.08651465054810563}, {"ibn_sina": 0.1462183513807652, "einstein": 0.02242331042811735, "darwin": 0.16568709867709575, "sagan": 0.06906865858920645, "hypatia": 0.45038422954405005, "socrates": 0.1462183513807652}, {"einstein": 0.01726384934589573, "sagan": 0.005604751286247881, "socrates": 0.009240672662620785, "ibn_sina": 0.32574493983354297, "darwin": 0.22388100476800438, "hypatia": 0.4182647821036883}, {"socrates": 0.021320014331370245, "ibn_sina": 0.1479902319873477, "hypatia": 0.10171209981069286, "einstein": 0.02914100449448594, "sagan": 0.2439946433333867, "darwin": 0.4558420060427166}], "prompt_tokens": 297, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.09905159933815753, "einstein": 0.013405176251469027, "hypatia": 0.20969223744439147, "ibn_sina": 0.5030255277420331, "sagan": 0.08741272961197438, "socrates": 0.08741272961197438}, "confidence": 0.22640719514043184, "choice": "ibn_sina"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 167.5238249990798, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "937427227f5f99118cb41f68bee040c493d9bec3da24346374497e87fd4b0c16", "compiled_ids_sha256": "f2a92c82350a676ec89ba5633141c479e87276520dc9842cd8d2d1356c426518", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.625, 15.625, 18.375, 19.25, 17.5, 17.5], "runtime_probabilities": [0.0990515947341919, 0.013405176810920238, 0.20969223976135254, 0.5030255317687988, 0.08741272985935211, 0.08741272985935211], "decide_seconds": 0.16360821499984013, "probabilities_f64_minus_runtime_max_abs": 4.60396563928267e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.11712047854482453, "hypatia": 0.13271488908551432, "sagan": 0.08049564923072598, "socrates": 0.10335845954503321, "darwin": 0.24794405502507794, "einstein": 0.318366468568824}, "confidence": 0.06977249100801708, "choice": "einstein"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.54911999992328, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "17b90d81647891d8db0520a988ad0cdae1e9f5bd1fed5dcd58e150aedbf9bc03", "compiled_ids_sha256": "ad9bc07ae62caa648cf4adc6a4b8d09b9940231b74b8f8c2faf04a6a55170a49", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 17.625, 17.125, 17.375, 18.25, 18.5], "runtime_probabilities": [0.11712049692869186, 0.1327148973941803, 0.08049565553665161, 0.10335846245288849, 0.24794407188892365, 0.31836649775505066], "decide_seconds": 0.161856758997601, "probabilities_f64_minus_runtime_max_abs": 2.9186226657706982e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.12587807378618757, "darwin": 0.028087194763087894, "sagan": 0.013267451366405308, "ibn_sina": 0.7243783002604417, "einstein": 0.021874329275771914, "socrates": 0.08651465054810563}, "confidence": 0.47119849086264187, "choice": "ibn_sina"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 214.0775379994011, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "0f3022a27181d5a0a95921e9f56219e94b652cc915de14ff442d6bba8f27fe29", "compiled_ids_sha256": "e6d851f1dddf7852b037b497b70026a76a6254e6f7ec9a64c0052f10a335a0de", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.875, 16.375, 15.625, 19.625, 16.125, 17.5], "runtime_probabilities": [0.12587806582450867, 0.02808719500899315, 0.013267451897263527, 0.7243782877922058, 0.021874329075217247, 0.08651465177536011], "decide_seconds": 0.2104680250013189, "probabilities_f64_minus_runtime_max_abs": 1.2468235888185575e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.1462183513807652, "einstein": 0.02242331042811735, "darwin": 0.16568709867709575, "sagan": 0.06906865858920645, "hypatia": 0.45038422954405005, "socrates": 0.1462183513807652}, "confidence": 0.16891396523888824, "choice": "hypatia"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.55882099783048, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "58f0b52eb9761ac361ed3c74d0385ea3ec690a657f7e3ee4907d149256845dab", "compiled_ids_sha256": "41784a82da1e97accdcff700414b1231236af276fc5ece9e41e0ae6ce5b977a8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.875, 16.0, 18.0, 17.125, 19.0, 17.875], "runtime_probabilities": [0.14621834456920624, 0.022423310205340385, 0.16568711400032043, 0.06906866282224655, 0.4503842294216156, 0.14621834456920624], "decide_seconds": 0.16165637000085553, "probabilities_f64_minus_runtime_max_abs": 1.532322468911218e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.01726384934589573, "sagan": 0.005604751286247881, "socrates": 0.009240672662620785, "ibn_sina": 0.32574493983354297, "darwin": 0.22388100476800438, "hypatia": 0.4182647821036883}, "confidence": 0.3261196299466782, "choice": "hypatia"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 166.39844399833237, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "3a18fbc6a84be5fcb500032684a23491003db5acc02f325e46b5f30aead3dee2", "compiled_ids_sha256": "a8a433667a748fb07b94519749f67325473a0289572d2acb178a434eeed3577a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 14.8125, 15.3125, 18.875, 18.5, 19.125], "runtime_probabilities": [0.017263850197196007, 0.005604751408100128, 0.009240671992301941, 0.3257449269294739, 0.22388100624084473, 0.418264776468277], "decide_seconds": 0.16277520099902176, "probabilities_f64_minus_runtime_max_abs": 1.2904069091046466e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.021320014331370245, "ibn_sina": 0.1479902319873477, "hypatia": 0.10171209981069286, "einstein": 0.02914100449448594, "sagan": 0.2439946433333867, "darwin": 0.4558420060427166}, "confidence": 0.21719787136343582, "choice": "darwin"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.45564100306365, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "dd944680d9992422a25d628a6e81a6b820f291626cc233ff3d8ba7afc2cc8b85", "compiled_ids_sha256": "0a7ed79ec519fae4996b7c78e47c1a5a5e9f9ca8dd751c48cd6d294f6d1cbee1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 17.875, 17.5, 16.25, 18.375, 19.0], "runtime_probabilities": [0.02132001519203186, 0.14799022674560547, 0.10171210020780563, 0.029141003265976906, 0.24399465322494507, 0.4558420181274414], "decide_seconds": 0.16184107799927006, "probabilities_f64_minus_runtime_max_abs": 1.2084724831940008e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:37Z"} {"id": "c-hypatia-01-main-01", "uid": "c042", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "prompt_sha256_by_order": ["79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60"], "request_sha256": "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "compiled_prompt_sha256_by_order": ["9e620d83c4565136473c23f396206defac1ffa16bd69dabc4a8fb76aa81893ae", "4ca5f41c1e88468acb0f861f5f2429161ef8d500bc13ba5dd73a19e226767655", "3cbbcc1ebea5302b168d6eccfe70024bcfe034f98c60ecb65abaf633d9aed5d2", "5c791f350bba1210c9a65b72356a61dc168f95d3e7f029aebef1e44ef4d75cc1", "a71462eb6cbb64abcd306c0d7907f55029c8e139b3bbc780467b01b369535509", "65fd84f15462d8af5cf515f76ed6c8b4f2a16c5d42201402c3045864dd35bf66"], "compiled_tokens_by_order": [388, 388, 388, 388, 388, 388], "choice": "socrates", "probs": {"darwin": 0.012501006946326287, "einstein": 0.023354955691306217, "hypatia": 0.23587669667413197, "ibn_sina": 0.12625569761613642, "sagan": 0.036172862997952755, "socrates": 0.5658387800741463}, "confidence": 0.3376336823653716, "seconds": 0.1791461076663836, "seconds_all_orders": 1.0748766459983017, "seconds_by_order": [0.18245785099861678, 0.17709256099988124, 0.1791577559997677, 0.1781857839996519, 0.18086926300020423, 0.17711343100017984], "argmax_stable": false, "spread": {"darwin": 0.09265588545713839, "einstein": 0.160818369812874, "hypatia": 0.4501917583147315, "ibn_sina": 0.18971373102409278, "sagan": 0.09420279304894788, "socrates": 0.5103478759349468}, "spread_2dp": {"darwin": 0.1, "einstein": 0.16, "hypatia": 0.46, "ibn_sina": 0.19, "sagan": 0.09, "socrates": 0.51}, "choices_by_order": ["socrates", "hypatia", "socrates", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.012501006946326287, "einstein": 0.023354955691306217, "hypatia": 0.23587669667413197, "ibn_sina": 0.12625569761613642, "sagan": 0.036172862997952755, "socrates": 0.5658387800741463}, {"ibn_sina": 0.07436947793536904, "hypatia": 0.3776793725321221, "sagan": 0.08427165887785087, "socrates": 0.2595749835618291, "darwin": 0.025701403848440885, "einstein": 0.17840310324438796}, {"hypatia": 0.1848626794856773, "darwin": 0.012580063216352072, "sagan": 0.0161531209133346, "ibn_sina": 0.1848626794856773, "einstein": 0.03212431681104185, "socrates": 0.5694171400879169}, {"ibn_sina": 0.04490418991356078, "einstein": 0.017584733431513965, "darwin": 0.004732866796467995, "sagan": 0.0740344930540458, "hypatia": 0.48276553787076876, "socrates": 0.3759781789336427}, {"einstein": 0.0205006854595263, "sagan": 0.01925861171730929, "socrates": 0.1423029623667215, "ibn_sina": 0.23461792093765357, "darwin": 0.0205006854595263, "hypatia": 0.5628191340592631}, {"socrates": 0.0590692641529701, "ibn_sina": 0.05212844265294739, "hypatia": 0.6350544378004088, "einstein": 0.046003189177784766, "sagan": 0.11035591396228248, "darwin": 0.09738875225360638}], "prompt_tokens": 388, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.012501006946326287, "einstein": 0.023354955691306217, "hypatia": 0.23587669667413197, "ibn_sina": 0.12625569761613642, "sagan": 0.036172862997952755, "socrates": 0.5658387800741463}, "confidence": 0.3376336823653716, "choice": "socrates"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 182.4296909981058, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "9e620d83c4565136473c23f396206defac1ffa16bd69dabc4a8fb76aa81893ae", "compiled_ids_sha256": "5cdc4054cab42c152b7033db49be91482544fff96b21837effc480511601a69b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.9375, 18.25, 17.625, 16.375, 19.125], "runtime_probabilities": [0.012501006945967674, 0.02335495874285698, 0.23587670922279358, 0.12625570595264435, 0.036172863095998764, 0.5658388137817383], "decide_seconds": 0.1778241270003491, "probabilities_f64_minus_runtime_max_abs": 3.370759193987993e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.07436947793536904, "hypatia": 0.3776793725321221, "sagan": 0.08427165887785087, "socrates": 0.2595749835618291, "darwin": 0.025701403848440885, "einstein": 0.17840310324438796}, "confidence": 0.15101158982647567, "choice": "hypatia"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 177.06632100089337, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "4ca5f41c1e88468acb0f861f5f2429161ef8d500bc13ba5dd73a19e226767655", "compiled_ids_sha256": "62c8a0a4a87a35dd1b0e0a2142b795c557227c4674a4b0759129b0090c13b730", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 18.375, 16.875, 18.0, 15.6875, 17.625], "runtime_probabilities": [0.07436947524547577, 0.37767937779426575, 0.08427166193723679, 0.2595750093460083, 0.025701403617858887, 0.1784031093120575], "decide_seconds": 0.17259510599978967, "probabilities_f64_minus_runtime_max_abs": 2.5784179180643463e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.1848626794856773, "darwin": 0.012580063216352072, "sagan": 0.0161531209133346, "ibn_sina": 0.1848626794856773, "einstein": 0.03212431681104185, "socrates": 0.5694171400879169}, "confidence": 0.3431331343693548, "choice": "socrates"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 179.13273600061075, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "3cbbcc1ebea5302b168d6eccfe70024bcfe034f98c60ecb65abaf633d9aed5d2", "compiled_ids_sha256": "cb010ed0292e1d26c5d8f62251685a997452f9d2a0e088f9600cc7333ca82088", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.25, 15.5625, 15.8125, 18.25, 16.5, 19.375], "runtime_probabilities": [0.18486268818378448, 0.012580063194036484, 0.0161531213670969, 0.18486268818378448, 0.032124318182468414, 0.5694171786308289], "decide_seconds": 0.17473689899998135, "probabilities_f64_minus_runtime_max_abs": 3.8542911973138416e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04490418991356078, "einstein": 0.017584733431513965, "darwin": 0.004732866796467995, "sagan": 0.0740344930540458, "hypatia": 0.48276553787076876, "socrates": 0.3759781789336427}, "confidence": 0.35938958302154356, "choice": "hypatia"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 178.15528399660252, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "5c791f350bba1210c9a65b72356a61dc168f95d3e7f029aebef1e44ef4d75cc1", "compiled_ids_sha256": "0f8d36ea8a5c218b7d2162d67f980a8151b2c6399f0597c968d3841fb62a5132", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 15.5625, 14.25, 17.0, 18.875, 18.625], "runtime_probabilities": [0.0449041910469532, 0.01758473552763462, 0.004732867237180471, 0.07403450459241867, 0.4827655851840973, 0.3759782016277313], "decide_seconds": 0.17360119799923268, "probabilities_f64_minus_runtime_max_abs": 4.731332853280534e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0205006854595263, "sagan": 0.01925861171730929, "socrates": 0.1423029623667215, "ibn_sina": 0.23461792093765357, "darwin": 0.0205006854595263, "hypatia": 0.5628191340592631}, "confidence": 0.34334430131763816, "choice": "hypatia"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 180.84395299956668, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "a71462eb6cbb64abcd306c0d7907f55029c8e139b3bbc780467b01b369535509", "compiled_ids_sha256": "be59331babc1741f9c6a119ab84993967e55001c780438b5c00f15bbc2552ce0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.75, 17.75, 18.25, 15.8125, 19.125], "runtime_probabilities": [0.02050068788230419, 0.019258612766861916, 0.1423029750585556, 0.2346179336309433, 0.02050068788230419, 0.5628191828727722], "decide_seconds": 0.17644975599978352, "probabilities_f64_minus_runtime_max_abs": 4.881350912366855e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0590692641529701, "ibn_sina": 0.05212844265294739, "hypatia": 0.6350544378004088, "einstein": 0.046003189177784766, "sagan": 0.11035591396228248, "darwin": 0.09738875225360638}, "confidence": 0.31846862460429604, "choice": "hypatia"}}, "usage": {"input_tokens": 388, "output_tokens": 1}, "latency_ms": 177.0879520008748, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 388, "compiled_prompt_sha256": "65fd84f15462d8af5cf515f76ed6c8b4f2a16c5d42201402c3045864dd35bf66", "compiled_ids_sha256": "72dd8f4c1e51c227ef6c4f5655c588ade075dd15eacbbc13317c7afd4a73b761", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 16.625, 19.125, 16.5, 17.375, 17.25], "runtime_probabilities": [0.05906926095485687, 0.052128441631793976, 0.6350544095039368, 0.04600318893790245, 0.110355906188488, 0.09738875180482864], "decide_seconds": 0.17271769500075607, "probabilities_f64_minus_runtime_max_abs": 2.8296472054378796e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:38Z"} {"id": "c-hypatia-01-main-01", "uid": "c043", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "prompt_sha256_by_order": ["1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568"], "request_sha256": "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "compiled_prompt_sha256_by_order": ["96b3447c5f6b1e9a91ad93d8847b48d6895167515fc8658b556b546cd722af75", "727f593328d2a66d56ff3b237949178a2b87a72508a3a53f2e7fe994cc222f35", "d5b11cdefbb4228224e0a6f9b37c43c8ade2eccc794c32841a45140b4935c41d", "3eaa0693bfbe41e644e724319a2e0700d21d42a1e292dfee99cd7bdda1af2c9e", "886ceda06e84376619e086e452d366d1d5adab0db0db140effd8d3e228b04c96", "8228536bbf2e08d2ec210e318ca04e418945ededf94ccd1d71c6e258bcd2735c"], "compiled_tokens_by_order": [339, 339, 339, 339, 339, 339], "choice": "socrates", "probs": {"darwin": 0.016364532705436224, "einstein": 0.018543444920326543, "hypatia": 0.165275925512073, "ibn_sina": 0.03254475936350589, "sagan": 0.11359237164670312, "socrates": 0.6536789658519552}, "confidence": 0.39990857738193364, "seconds": 0.17355502716600313, "seconds_all_orders": 1.0413301629960188, "seconds_by_order": [0.17937791399890557, 0.17083098999864887, 0.17088872899694252, 0.17077110100217396, 0.17709193200062145, 0.1723694969987264], "argmax_stable": false, "spread": {"darwin": 0.051394905222396856, "einstein": 0.12048500522872077, "hypatia": 0.3759439738857655, "ibn_sina": 0.03476666191853326, "sagan": 0.14014369709770952, "socrates": 0.3754474630785295}, "spread_2dp": {"darwin": 0.049999999999999996, "einstein": 0.12000000000000001, "hypatia": 0.38, "ibn_sina": 0.04000000000000001, "sagan": 0.14, "socrates": 0.38000000000000006}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.016364532705436224, "einstein": 0.018543444920326543, "hypatia": 0.165275925512073, "ibn_sina": 0.03254475936350589, "sagan": 0.11359237164670312, "socrates": 0.6536789658519552}, {"ibn_sina": 0.05050496016502012, "hypatia": 0.29063597891718757, "sagan": 0.09435568765857426, "socrates": 0.4228728541036645, "darwin": 0.03471151764718346, "einstein": 0.10691900150837005}, {"hypatia": 0.06298162902675064, "darwin": 0.008007215567569509, "sagan": 0.05558109253583948, "ibn_sina": 0.06298162902675064, "einstein": 0.1333321096959244, "socrates": 0.6771163241471654}, {"ibn_sina": 0.034922050621514676, "einstein": 0.01284710446720364, "darwin": 0.010005335019258337, "sagan": 0.030818601505389148, "hypatia": 0.2923987499291334, "socrates": 0.6190081584575008}, {"einstein": 0.08642957575951443, "sagan": 0.04626241818732798, "socrates": 0.3016688610686359, "ibn_sina": 0.06731142128203915, "darwin": 0.059402120789966366, "hypatia": 0.43892560291251614}, {"socrates": 0.3193996232385596, "ibn_sina": 0.04898151863203195, "hypatia": 0.36192718898290094, "einstein": 0.055503332066750934, "sagan": 0.17096229860309867, "darwin": 0.043226038476657846}], "prompt_tokens": 339, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.016364532705436224, "einstein": 0.018543444920326543, "hypatia": 0.165275925512073, "ibn_sina": 0.03254475936350589, "sagan": 0.11359237164670312, "socrates": 0.6536789658519552}, "confidence": 0.39990857738193364, "choice": "socrates"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 179.35047400169424, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "96b3447c5f6b1e9a91ad93d8847b48d6895167515fc8658b556b546cd722af75", "compiled_ids_sha256": "a5315fe99a572917cd0eab60dbea5014cc09871c3719dbb84d88bd7901126ebf", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.6875, 17.875, 16.25, 17.5, 19.25], "runtime_probabilities": [0.016364535316824913, 0.01854344643652439, 0.1652759462594986, 0.03254476189613342, 0.11359237134456635, 0.6536790132522583], "decide_seconds": 0.1752360860009503, "probabilities_f64_minus_runtime_max_abs": 4.740030312699872e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.05050496016502012, "hypatia": 0.29063597891718757, "sagan": 0.09435568765857426, "socrates": 0.4228728541036645, "darwin": 0.03471151764718346, "einstein": 0.10691900150837005}, "confidence": 0.18944409866306577, "choice": "socrates"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.80708999856142, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "727f593328d2a66d56ff3b237949178a2b87a72508a3a53f2e7fe994cc222f35", "compiled_ids_sha256": "fe54e2af12285f2a723f14ddc1a2bd915bb9e4b9d0aa5b734019194b4aac6475", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 18.5, 17.375, 18.875, 16.375, 17.5], "runtime_probabilities": [0.050504960119724274, 0.29063597321510315, 0.09435568749904633, 0.4228728413581848, 0.03471151739358902, 0.10691899806261063], "decide_seconds": 0.16667184100151644, "probabilities_f64_minus_runtime_max_abs": 1.2745479671671944e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.06298162902675064, "darwin": 0.008007215567569509, "sagan": 0.05558109253583948, "ibn_sina": 0.06298162902675064, "einstein": 0.1333321096959244, "socrates": 0.6771163241471654}, "confidence": 0.3971153031584267, "choice": "socrates"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.86533999827225, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "d5b11cdefbb4228224e0a6f9b37c43c8ade2eccc794c32841a45140b4935c41d", "compiled_ids_sha256": "23fbfc222e8036a868893934c7aa06b92cfb2f659933da08dad21e131ac6da46", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 14.9375, 16.875, 17.0, 17.75, 19.375], "runtime_probabilities": [0.06298163533210754, 0.008007216267287731, 0.055581096559762955, 0.06298163533210754, 0.13333211839199066, 0.6771163940429688], "decide_seconds": 0.16692140899976948, "probabilities_f64_minus_runtime_max_abs": 6.989580336735912e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.034922050621514676, "einstein": 0.01284710446720364, "darwin": 0.010005335019258337, "sagan": 0.030818601505389148, "hypatia": 0.2923987499291334, "socrates": 0.6190081584575008}, "confidence": 0.4514625749336868, "choice": "socrates"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.74769099781406, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "3eaa0693bfbe41e644e724319a2e0700d21d42a1e292dfee99cd7bdda1af2c9e", "compiled_ids_sha256": "232e1dc5fe06f394111fd1f330f33a99b1e2babef926933a554298012f04b014", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 15.375, 15.125, 16.25, 18.5, 19.25], "runtime_probabilities": [0.034922052174806595, 0.012847104109823704, 0.010005335323512554, 0.030818600207567215, 0.29239875078201294, 0.6190081834793091], "decide_seconds": 0.16673025100317318, "probabilities_f64_minus_runtime_max_abs": 2.5021808292002845e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.08642957575951443, "sagan": 0.04626241818732798, "socrates": 0.3016688610686359, "ibn_sina": 0.06731142128203915, "darwin": 0.059402120789966366, "hypatia": 0.43892560291251614}, "confidence": 0.2040768556698579, "choice": "hypatia"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 177.06875200019567, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "886ceda06e84376619e086e452d366d1d5adab0db0db140effd8d3e228b04c96", "compiled_ids_sha256": "4e2f2e110205e1684b6808fe1ab6b3a09caa33a2ca0a3209142a652ad8eef405", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 16.625, 18.5, 17.0, 16.875, 18.875], "runtime_probabilities": [0.08642956614494324, 0.04626241326332092, 0.30166885256767273, 0.06731142103672028, 0.05940211936831474, 0.4389255940914154], "decide_seconds": 0.17312529199989513, "probabilities_f64_minus_runtime_max_abs": 9.614571194749288e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.3193996232385596, "ibn_sina": 0.04898151863203195, "hypatia": 0.36192718898290094, "einstein": 0.055503332066750934, "sagan": 0.17096229860309867, "darwin": 0.043226038476657846}, "confidence": 0.1749193785819163, "choice": "hypatia"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 172.3451580000983, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "8228536bbf2e08d2ec210e318ca04e418945ededf94ccd1d71c6e258bcd2735c", "compiled_ids_sha256": "48b0351d721ebeacce205e0e28dcd105814af68abcf16e166271126420fa4fa4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.875, 17.0, 19.0, 17.125, 18.25, 16.875], "runtime_probabilities": [0.3193996250629425, 0.04898151755332947, 0.36192718148231506, 0.05550333485007286, 0.17096230387687683, 0.04322604089975357], "decide_seconds": 0.16832462899765233, "probabilities_f64_minus_runtime_max_abs": 7.500585874620214e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:39Z"} {"id": "c-hypatia-01-main-01", "uid": "c044", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "prompt_sha256_by_order": ["550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0"], "request_sha256": "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "compiled_prompt_sha256_by_order": ["12910ec5c352e9fa1330272cb09c7a1e3ca9ef566bd28df789137a8794a156bc", "807d8a79dbcf0eff03407ca148654a28f33c941d1b260bbbe9deb3edbed4c903", "3775085ccc872c9b384b145b06b80f3a2a0cb2deb35c9b642e3d16edf2c1a06b", "dddb54acc96be195aa3e9ab1c24a8384413f94d3b46ba7004698c660b361c35c", "e414d533753b06ef4a940fdcc038fbe8a9bff466fd2a6493da0716b9f484077f", "d2b15f3e99a322a75aff55d2228be720fa0786b4ffb917d8ca30879880ae7d96"], "compiled_tokens_by_order": [275, 275, 275, 275, 275, 275], "choice": "sagan", "probs": {"darwin": 0.013001924016967176, "einstein": 0.13978385655077696, "hypatia": 0.07482090674126249, "ibn_sina": 0.006141674026895391, "sagan": 0.626467782113321, "socrates": 0.13978385655077696}, "confidence": 0.3722418411056071, "seconds": 0.16762706433291896, "seconds_all_orders": 1.0057623859975138, "seconds_by_order": [0.16577698800028884, 0.16520405299888807, 0.17123998700117227, 0.16986698699838598, 0.16699269899982028, 0.16668167199895834], "argmax_stable": false, "spread": {"darwin": 0.023036583641724707, "einstein": 0.6182091322829826, "hypatia": 0.09306361889554943, "ibn_sina": 0.013028719279082291, "sagan": 0.6131828596842024, "socrates": 0.13135356218891037}, "spread_2dp": {"darwin": 0.019999999999999997, "einstein": 0.62, "hypatia": 0.09000000000000001, "ibn_sina": 0.02, "sagan": 0.6200000000000001, "socrates": 0.13}, "choices_by_order": ["sagan", "einstein", "sagan", "sagan", "einstein", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.013001924016967176, "einstein": 0.13978385655077696, "hypatia": 0.07482090674126249, "ibn_sina": 0.006141674026895391, "sagan": 0.626467782113321, "socrates": 0.13978385655077696}, {"ibn_sina": 0.008053095860962143, "hypatia": 0.007565183447876973, "sagan": 0.3021901927603331, "socrates": 0.021890584141692145, "darwin": 0.02056430069532312, "einstein": 0.6397366430938125}, {"hypatia": 0.01775418401938613, "darwin": 0.009503129900405648, "sagan": 0.85544449201073, "ibn_sina": 0.0044889607113046785, "einstein": 0.07021915994952259, "socrates": 0.042590073408650855}, {"ibn_sina": 0.016330069985730947, "einstein": 0.05354423810481588, "darwin": 0.014411236181397231, "sagan": 0.7391554071543714, "hypatia": 0.08827952428684226, "socrates": 0.08827952428684226}, {"einstein": 0.3979946338038614, "sagan": 0.351229031577198, "socrates": 0.11402737170942105, "ibn_sina": 0.007289525992405522, "darwin": 0.0288306345736876, "hypatia": 0.1006288023434264}, {"socrates": 0.008430294361866602, "ibn_sina": 0.0033013507066486556, "hypatia": 0.0455737407441564, "einstein": 0.021527510810829882, "sagan": 0.9153730524445355, "darwin": 0.005794050931962896}], "prompt_tokens": 275, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.013001924016967176, "einstein": 0.13978385655077696, "hypatia": 0.07482090674126249, "ibn_sina": 0.006141674026895391, "sagan": 0.626467782113321, "socrates": 0.13978385655077696}, "confidence": 0.3722418411056071, "choice": "sagan"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 165.75083800125867, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "12910ec5c352e9fa1330272cb09c7a1e3ca9ef566bd28df789137a8794a156bc", "compiled_ids_sha256": "bf822f6c8b673ada113561218d839770895abe054bccedd183916316a0e84576", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 17.75, 17.125, 14.625, 19.25, 17.75], "runtime_probabilities": [0.013001924380660057, 0.1397838592529297, 0.07482091337442398, 0.00614167470484972, 0.6264678239822388, 0.1397838592529297], "decide_seconds": 0.16222500599906198, "probabilities_f64_minus_runtime_max_abs": 4.186891777724355e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.008053095860962143, "hypatia": 0.007565183447876973, "sagan": 0.3021901927603331, "socrates": 0.021890584141692145, "darwin": 0.02056430069532312, "einstein": 0.6397366430938125}, "confidence": 0.5051151433322534, "choice": "einstein"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 165.18053300023894, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "807d8a79dbcf0eff03407ca148654a28f33c941d1b260bbbe9deb3edbed4c903", "compiled_ids_sha256": "24d303cadcd22aa980d79c46d494051003717dd886114c4507fb2524aa82a045", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 14.8125, 18.5, 15.875, 15.8125, 19.25], "runtime_probabilities": [0.008053096011281013, 0.007565183099359274, 0.3021901845932007, 0.021890582516789436, 0.020564300939440727, 0.6397366523742676], "decide_seconds": 0.16167909000068903, "probabilities_f64_minus_runtime_max_abs": 9.28045507109232e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.01775418401938613, "darwin": 0.009503129900405648, "sagan": 0.85544449201073, "ibn_sina": 0.0044889607113046785, "einstein": 0.07021915994952259, "socrates": 0.042590073408650855}, "confidence": 0.6681581467219209, "choice": "sagan"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 171.21718699854682, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "3775085ccc872c9b384b145b06b80f3a2a0cb2deb35c9b642e3d16edf2c1a06b", "compiled_ids_sha256": "73cdc93d43cefdf705b52bb4cf5773efb4c2b9e507e3baf50004a8f49463a85e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 15.375, 19.875, 14.625, 17.375, 16.875], "runtime_probabilities": [0.01775418221950531, 0.009503129869699478, 0.8554444313049316, 0.0044889603741467, 0.07021915912628174, 0.04259007051587105], "decide_seconds": 0.1678490519989282, "probabilities_f64_minus_runtime_max_abs": 6.070579838812762e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.016330069985730947, "einstein": 0.05354423810481588, "darwin": 0.014411236181397231, "sagan": 0.7391554071543714, "hypatia": 0.08827952428684226, "socrates": 0.08827952428684226}, "confidence": 0.477055185188223, "choice": "sagan"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 169.8438970015559, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "dddb54acc96be195aa3e9ab1c24a8384413f94d3b46ba7004698c660b361c35c", "compiled_ids_sha256": "58d50768ea1738f96ec68f9834c8140601d064cdc382fff1ca4128c75457b6aa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 17.0, 15.6875, 19.625, 17.5, 17.5], "runtime_probabilities": [0.016330068930983543, 0.053544238209724426, 0.014411237090826035, 0.7391554117202759, 0.08827952295541763, 0.08827952295541763], "decide_seconds": 0.16633218400238547, "probabilities_f64_minus_runtime_max_abs": 4.565904432318746e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.3979946338038614, "sagan": 0.351229031577198, "socrates": 0.11402737170942105, "ibn_sina": 0.007289525992405522, "darwin": 0.0288306345736876, "hypatia": 0.1006288023434264}, "confidence": 0.24601613864656235, "choice": "einstein"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 166.9697199977236, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "e414d533753b06ef4a940fdcc038fbe8a9bff466fd2a6493da0716b9f484077f", "compiled_ids_sha256": "270b5d8a6c423fdc32fb16d3c6ace64acc0368b85487141cfc4f2f40250e5fb7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.875, 18.75, 17.625, 14.875, 16.25, 17.5], "runtime_probabilities": [0.39799460768699646, 0.3512290120124817, 0.11402736604213715, 0.0072895255871117115, 0.02883063443005085, 0.1006287932395935], "decide_seconds": 0.1635675750003429, "probabilities_f64_minus_runtime_max_abs": 2.6116864948466656e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.008430294361866602, "ibn_sina": 0.0033013507066486556, "hypatia": 0.0455737407441564, "einstein": 0.021527510810829882, "sagan": 0.9153730524445355, "darwin": 0.005794050931962896}, "confidence": 0.7804992974091274, "choice": "sagan"}}, "usage": {"input_tokens": 275, "output_tokens": 1}, "latency_ms": 166.65940199891338, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 275, "compiled_prompt_sha256": "d2b15f3e99a322a75aff55d2228be720fa0786b4ffb917d8ca30879880ae7d96", "compiled_ids_sha256": "ddb453eb0bdf05edbcd3f6e05e44b2cc37bfd387ac0ba8e59a25ed59290ad899", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 14.0, 16.625, 15.875, 19.625, 14.5625], "runtime_probabilities": [0.008430294692516327, 0.0033013508655130863, 0.04557374119758606, 0.02152751199901104, 0.9153730869293213, 0.005794051103293896], "decide_seconds": 0.16327892700064695, "probabilities_f64_minus_runtime_max_abs": 3.448478580470038e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:40Z"} {"id": "c-hypatia-01-main-02", "uid": "c045", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "prompt_sha256_by_order": ["a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2"], "request_sha256": "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "compiled_prompt_sha256_by_order": ["53ea90fe5d10904b59900dd3288e3f69d354a8b53be8832d2ed7310f00a05a78", "7b8ba837eec9ce754951db3caac4fd2da2ca873b5a41214f681f3059bf74da7a", "4f5ec06b6ec67c809734ca47b5b836ce712df5bf1d4eed101c7b9f244a430910", "dc1f71e89449ba9f6ec5ec75538f0ac800608cc1bf7eabab422b36013ac66370", "56e7f69d049c3c61a0c93373197985ba905699c9fdca194d77cb1dc097f03228", "5af5f17bf2c762b293eb16ba8999f89bbfdbe8db14a42dd8732b5bafceca03b6"], "compiled_tokens_by_order": [349, 349, 349, 349, 349, 349], "choice": "socrates", "probs": {"darwin": 0.014368062196211943, "einstein": 0.1280611443826663, "hypatia": 0.1280611443826663, "ibn_sina": 0.18632786562296139, "sagan": 0.036690132157039285, "socrates": 0.5064916512584547}, "confidence": 0.23748622889607762, "seconds": 0.17157940033333338, "seconds_all_orders": 1.0294764020000002, "seconds_by_order": [0.17170020300181932, 0.17128262600090238, 0.17095936799887568, 0.17110843699992984, 0.170692080999288, 0.173733686999185], "argmax_stable": false, "spread": {"darwin": 0.06786034261041284, "einstein": 0.33521898487019813, "hypatia": 0.3047303300633457, "ibn_sina": 0.15380964791756002, "sagan": 0.14963255277202758, "socrates": 0.5359838252252462}, "spread_2dp": {"darwin": 0.07, "einstein": 0.33999999999999997, "hypatia": 0.3, "ibn_sina": 0.16, "sagan": 0.15, "socrates": 0.54}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.014368062196211943, "einstein": 0.1280611443826663, "hypatia": 0.1280611443826663, "ibn_sina": 0.18632786562296139, "sagan": 0.036690132157039285, "socrates": 0.5064916512584547}, {"ibn_sina": 0.036396981993773746, "hypatia": 0.12703794977483365, "sagan": 0.011100452341187688, "socrates": 0.6451519425808422, "darwin": 0.017192716914575686, "einstein": 0.16311995639478713}, {"hypatia": 0.13528295658146103, "darwin": 0.007169734033882919, "sagan": 0.009206120730396795, "ibn_sina": 0.06390314384499711, "einstein": 0.3677372025755995, "socrates": 0.41670084223366266}, {"ibn_sina": 0.03251821770540135, "einstein": 0.03251821770540135, "darwin": 0.012734317022936278, "sagan": 0.017405747661884647, "hypatia": 0.39615299080889277, "socrates": 0.5086705090954835}, {"einstein": 0.08351225849583982, "sagan": 0.016444538720501507, "socrates": 0.3302973896554223, "ibn_sina": 0.15602143933088752, "darwin": 0.03944839765729637, "hypatia": 0.37427597614005254}, {"socrates": 0.10916811735559594, "ibn_sina": 0.08502021528296916, "hypatia": 0.4317682798381794, "einstein": 0.14017463737653535, "sagan": 0.15883867350242437, "darwin": 0.07503007664429576}], "prompt_tokens": 349, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.014368062196211943, "einstein": 0.1280611443826663, "hypatia": 0.1280611443826663, "ibn_sina": 0.18632786562296139, "sagan": 0.036690132157039285, "socrates": 0.5064916512584547}, "confidence": 0.23748622889607762, "choice": "socrates"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.67349300143542, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "53ea90fe5d10904b59900dd3288e3f69d354a8b53be8832d2ed7310f00a05a78", "compiled_ids_sha256": "fc14aba23487d6da377ebf5043577838bb63cffc65b91089be20b810df6d2bad", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 17.625, 17.625, 18.0, 16.375, 19.0], "runtime_probabilities": [0.01436806283891201, 0.12806114554405212, 0.12806114554405212, 0.18632788956165314, 0.036690134555101395, 0.5064916610717773], "decide_seconds": 0.16733259600005113, "probabilities_f64_minus_runtime_max_abs": 2.393869175110197e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.036396981993773746, "hypatia": 0.12703794977483365, "sagan": 0.011100452341187688, "socrates": 0.6451519425808422, "darwin": 0.017192716914575686, "einstein": 0.16311995639478713}, "confidence": 0.3966508581578657, "choice": "socrates"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.25614599717665, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "7b8ba837eec9ce754951db3caac4fd2da2ca873b5a41214f681f3059bf74da7a", "compiled_ids_sha256": "9a912b27d0a201b4913a165776dfd88073638d05b912990a61ae5735aa3462d6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 17.5, 15.0625, 19.125, 15.5, 17.75], "runtime_probabilities": [0.03639698401093483, 0.12703794240951538, 0.011100451461970806, 0.6451519131660461, 0.017192715778946877, 0.16311995685100555], "decide_seconds": 0.16694372900019516, "probabilities_f64_minus_runtime_max_abs": 2.9414796043347735e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.13528295658146103, "darwin": 0.007169734033882919, "sagan": 0.009206120730396795, "ibn_sina": 0.06390314384499711, "einstein": 0.3677372025755995, "socrates": 0.41670084223366266}, "confidence": 0.29812511559867094, "choice": "socrates"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 170.93554899838637, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "4f5ec06b6ec67c809734ca47b5b836ce712df5bf1d4eed101c7b9f244a430910", "compiled_ids_sha256": "00d68204ee82e667e49ffce720b827a5438f82b83114a757fb965a9142f3e1de", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.875, 14.9375, 15.1875, 17.125, 18.875, 19.0], "runtime_probabilities": [0.1352829486131668, 0.007169733755290508, 0.009206120856106281, 0.06390314549207687, 0.36773720383644104, 0.4167008399963379], "decide_seconds": 0.16683059099887032, "probabilities_f64_minus_runtime_max_abs": 7.968294218896332e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03251821770540135, "einstein": 0.03251821770540135, "darwin": 0.012734317022936278, "sagan": 0.017405747661884647, "hypatia": 0.39615299080889277, "socrates": 0.5086705090954835}, "confidence": 0.40865643097130677, "choice": "socrates"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 171.08396800176706, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "dc1f71e89449ba9f6ec5ec75538f0ac800608cc1bf7eabab422b36013ac66370", "compiled_ids_sha256": "f2547ae4afc2cc6b367dd1a27eba35253a4b8f94069ad1981f0f2976a1e99fd8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 16.25, 15.3125, 15.625, 18.75, 19.0], "runtime_probabilities": [0.03251821920275688, 0.03251821920275688, 0.012734318152070045, 0.01740574836730957, 0.3961530327796936, 0.5086705684661865], "decide_seconds": 0.16692465000232914, "probabilities_f64_minus_runtime_max_abs": 5.937070302053371e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.08351225849583982, "sagan": 0.016444538720501507, "socrates": 0.3302973896554223, "ibn_sina": 0.15602143933088752, "darwin": 0.03944839765729637, "hypatia": 0.37427597614005254}, "confidence": 0.20414303752653762, "choice": "hypatia"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 170.668320999539, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "56e7f69d049c3c61a0c93373197985ba905699c9fdca194d77cb1dc097f03228", "compiled_ids_sha256": "c54d8a0a1ed8ffd2b1ef1fd264e0e0bcbb3f064db7206763949f88f42737c4b9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 15.875, 18.875, 18.125, 16.75, 19.0], "runtime_probabilities": [0.08351225405931473, 0.016444537788629532, 0.3302973508834839, 0.15602143108844757, 0.03944839537143707, 0.37427595257759094], "decide_seconds": 0.16653480199965998, "probabilities_f64_minus_runtime_max_abs": 3.877193843759841e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.10916811735559594, "ibn_sina": 0.08502021528296916, "hypatia": 0.4317682798381794, "einstein": 0.14017463737653535, "sagan": 0.15883867350242437, "darwin": 0.07503007664429576}, "confidence": 0.12043550663305658, "choice": "hypatia"}}, "usage": {"input_tokens": 349, "output_tokens": 1}, "latency_ms": 173.7096870019741, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 349, "compiled_prompt_sha256": "5af5f17bf2c762b293eb16ba8999f89bbfdbe8db14a42dd8732b5bafceca03b6", "compiled_ids_sha256": "f906c10057f7a5b490afce1284e09e974c3469d23ce1191b29e087bb35b9907e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 17.25, 18.875, 17.75, 17.875, 17.125], "runtime_probabilities": [0.10916811227798462, 0.08502020686864853, 0.4317682683467865, 0.14017462730407715, 0.15883868932724, 0.07503006607294083], "decide_seconds": 0.16962579900064156, "probabilities_f64_minus_runtime_max_abs": 1.5824815624565503e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:41Z"} {"id": "c-hypatia-01-main-03", "uid": "c046", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "prompt_sha256_by_order": ["51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe"], "request_sha256": "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "compiled_prompt_sha256_by_order": ["981f394b92cdbfe83d1629e86ff1043844d3ddec1aa960eee15412baa3b51c00", "214a087a51941e97f1c28690b67a09812e8d200430880e5f296a79504f2dfc50", "da154d0f83a59d3e823b138fe84d86d01b16328c89c63f73015febd9cabf2e5e", "81999bb5b2c7b6db7853aa583b6e07a90b67369699ec5fb2dc1fdd51e0c21d22", "532d14b926253ef4b4304a2a1f635221b18243ab97d9c594080fc80b1dcb97fc", "d725197c1e2a979cdabd3217a3603d86ad6a186762c6e489efbb13604dcf97d6"], "compiled_tokens_by_order": [364, 364, 364, 364, 364, 364], "choice": "hypatia", "probs": {"darwin": 0.01063505363782502, "einstein": 0.016471893301862554, "hypatia": 0.6579674698323895, "ibn_sina": 0.025512167421793353, "sagan": 0.10090257952118757, "socrates": 0.188510836284942}, "confidence": 0.42461310604728764, "seconds": 0.17562262266740922, "seconds_all_orders": 1.0537357360044552, "seconds_by_order": [0.17375570700096432, 0.1720788500024355, 0.17575045200283057, 0.17893997699866304, 0.1763453470011882, 0.1768654029983736], "argmax_stable": true, "spread": {"darwin": 0.06548643197447994, "einstein": 0.08453767509423946, "hypatia": 0.22314905699233933, "ibn_sina": 0.04884560754216051, "sagan": 0.20044847039447009, "socrates": 0.2116909572777489}, "spread_2dp": {"darwin": 0.07, "einstein": 0.09000000000000001, "hypatia": 0.21999999999999997, "ibn_sina": 0.05, "sagan": 0.2, "socrates": 0.21000000000000002}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01063505363782502, "einstein": 0.016471893301862554, "hypatia": 0.6579674698323895, "ibn_sina": 0.025512167421793353, "sagan": 0.10090257952118757, "socrates": 0.188510836284942}, {"ibn_sina": 0.03402221215322316, "hypatia": 0.7743419794570691, "sagan": 0.03402221215322316, "socrates": 0.049502026589258716, "darwin": 0.026496525466751658, "einstein": 0.08161504418047416}, {"hypatia": 0.5511929224647297, "darwin": 0.012177437805861521, "sagan": 0.045244670282020095, "ibn_sina": 0.06583060681757051, "einstein": 0.09578296773867152, "socrates": 0.2297713948911467}, {"ibn_sina": 0.025341658166620477, "einstein": 0.011245292644432069, "darwin": 0.022363934838400188, "sagan": 0.10022820233501081, "hypatia": 0.6535699781824537, "socrates": 0.18725093383308278}, {"einstein": 0.04058089346441385, "sagan": 0.18187094670401696, "socrates": 0.052106898640204846, "ibn_sina": 0.03160443160782277, "darwin": 0.059044851588258035, "hypatia": 0.6347919779952835}, {"socrates": 0.018080437613397806, "ibn_sina": 0.01698499927541, "hypatia": 0.637357395675784, "einstein": 0.01698499927541, "sagan": 0.23447068254769324, "darwin": 0.07612148561230496}], "prompt_tokens": 364, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01063505363782502, "einstein": 0.016471893301862554, "hypatia": 0.6579674698323895, "ibn_sina": 0.025512167421793353, "sagan": 0.10090257952118757, "socrates": 0.188510836284942}, "confidence": 0.42461310604728764, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 173.72900799819035, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "981f394b92cdbfe83d1629e86ff1043844d3ddec1aa960eee15412baa3b51c00", "compiled_ids_sha256": "e1be0ad9a565296219335b76cdbbdeef9cf60282395d86cca26940dc6907053b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 15.9375, 19.625, 16.375, 17.75, 18.375], "runtime_probabilities": [0.010635052807629108, 0.016471894457936287, 0.6579674482345581, 0.02551216632127762, 0.10090257972478867, 0.18851083517074585], "decide_seconds": 0.16939529000228504, "probabilities_f64_minus_runtime_max_abs": 2.1597831412201174e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03402221215322316, "hypatia": 0.7743419794570691, "sagan": 0.03402221215322316, "socrates": 0.049502026589258716, "darwin": 0.026496525466751658, "einstein": 0.08161504418047416}, "confidence": 0.5102184718125369, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 172.05398000078276, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "214a087a51941e97f1c28690b67a09812e8d200430880e5f296a79504f2dfc50", "compiled_ids_sha256": "b5d6e2a37517c392d89a5125488f28128d5b96ea05775c430855018536e03c59", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 19.375, 16.25, 16.625, 16.0, 17.125], "runtime_probabilities": [0.03402221202850342, 0.7743419408798218, 0.03402221202850342, 0.04950202256441116, 0.02649652399122715, 0.08161503821611404], "decide_seconds": 0.16781562300093356, "probabilities_f64_minus_runtime_max_abs": 3.8577247285509486e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.5511929224647297, "darwin": 0.012177437805861521, "sagan": 0.045244670282020095, "ibn_sina": 0.06583060681757051, "einstein": 0.09578296773867152, "socrates": 0.2297713948911467}, "confidence": 0.2946772252638603, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 175.7252019997395, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "da154d0f83a59d3e823b138fe84d86d01b16328c89c63f73015febd9cabf2e5e", "compiled_ids_sha256": "2cb0828d9884ce5b5b4b7bb65767edfb2d72c7c8f63be63b02de221c96abfddd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 15.4375, 16.75, 17.125, 17.5, 18.375], "runtime_probabilities": [0.5511929392814636, 0.012177437543869019, 0.045244671404361725, 0.06583061069250107, 0.09578296542167664, 0.22977139055728912], "decide_seconds": 0.17135394499928225, "probabilities_f64_minus_runtime_max_abs": 1.6816733894309266e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.025341658166620477, "einstein": 0.011245292644432069, "darwin": 0.022363934838400188, "sagan": 0.10022820233501081, "hypatia": 0.6535699781824537, "socrates": 0.18725093383308278}, "confidence": 0.413526254661835, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 178.9148980024038, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "81999bb5b2c7b6db7853aa583b6e07a90b67369699ec5fb2dc1fdd51e0c21d22", "compiled_ids_sha256": "97ed5bb9478331d4df11428579ccac8270e79aab1730f49c36a0bdd0086fe500", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 15.4375, 16.125, 17.625, 19.5, 18.25], "runtime_probabilities": [0.025341657921671867, 0.011245292611420155, 0.02236393466591835, 0.1002282053232193, 0.6535699963569641, 0.18725094199180603], "decide_seconds": 0.17468228999860003, "probabilities_f64_minus_runtime_max_abs": 1.817451045216245e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.04058089346441385, "sagan": 0.18187094670401696, "socrates": 0.052106898640204846, "ibn_sina": 0.03160443160782277, "darwin": 0.059044851588258035, "hypatia": 0.6347919779952835}, "confidence": 0.3533137947803564, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 176.31948700000066, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "532d14b926253ef4b4304a2a1f635221b18243ab97d9c594080fc80b1dcb97fc", "compiled_ids_sha256": "3ce4781acdffa8bac9ded1b451485eb8cd5da890f2f1df0aaa4feb47d745f7b6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 18.125, 16.875, 16.375, 17.0, 19.375], "runtime_probabilities": [0.04058089479804039, 0.1818709522485733, 0.05210689827799797, 0.03160443156957626, 0.05904485285282135, 0.6347919702529907], "decide_seconds": 0.1720548110024538, "probabilities_f64_minus_runtime_max_abs": 7.742292806867113e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.018080437613397806, "ibn_sina": 0.01698499927541, "hypatia": 0.637357395675784, "einstein": 0.01698499927541, "sagan": 0.23447068254769324, "darwin": 0.07612148561230496}, "confidence": 0.4227984386241114, "choice": "hypatia"}}, "usage": {"input_tokens": 364, "output_tokens": 1}, "latency_ms": 176.83951299841283, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 364, "compiled_prompt_sha256": "d725197c1e2a979cdabd3217a3603d86ad6a186762c6e489efbb13604dcf97d6", "compiled_ids_sha256": "f1dc423f6d5a6e2d18d92dc8aa688fe9ac56e2c0166b826629d31f4318efa327", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.75, 19.375, 15.75, 18.375, 17.25], "runtime_probabilities": [0.018080437555909157, 0.016984999179840088, 0.6373574137687683, 0.016984999179840088, 0.23447071015834808, 0.076121486723423], "decide_seconds": 0.17256129700035672, "probabilities_f64_minus_runtime_max_abs": 2.7610654845799942e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:42Z"} {"id": "c-hypatia-01-main-03", "uid": "c047", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "prompt_sha256_by_order": ["ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f"], "request_sha256": "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "compiled_prompt_sha256_by_order": ["5e16fbf7126da6744631cdedb3c9e498a2b9052c185376919d00b1f279b12f9c", "8a9783359ca22b2ca060f289690fe6ed73f55643bd619baf4637a1b065c7ecca", "9878e42d49f4c5957122bb2b720a9531598548b6d8d8120ede38748a5a1a24de", "eb5b9290878f211609814ff1f2be0464ce38ec64a9a7a6beac89e8e9aa525329", "4440f5d875c123e4ff6fb95694a3ffa472d0c6be75bddc484d62adea6dfb75bb", "977d0b133c97885600b1c793fd8b912d7fdad5da8c8e2667e501ef1416aeaa88"], "compiled_tokens_by_order": [329, 329, 329, 329, 329, 329], "choice": "sagan", "probs": {"darwin": 0.01661376493972915, "einstein": 0.04242473500882781, "hypatia": 0.148076875060247, "ibn_sina": 0.00784778687262939, "sagan": 0.7519964212720947, "socrates": 0.033040416846471944}, "confidence": 0.5255967261252359, "seconds": 0.1730952886664454, "seconds_all_orders": 1.0385717319986725, "seconds_by_order": [0.17128091599806794, 0.17206140000052983, 0.17116266700031701, 0.17508185699989554, 0.17096949800179573, 0.17801539399806643], "argmax_stable": true, "spread": {"darwin": 0.02653639081055991, "einstein": 0.24692266985880978, "hypatia": 0.27256916739900966, "ibn_sina": 0.018734542317285344, "sagan": 0.19310784711003748, "socrates": 0.1019557897641788}, "spread_2dp": {"darwin": 0.03, "einstein": 0.24000000000000002, "hypatia": 0.27, "ibn_sina": 0.019999999999999997, "sagan": 0.18999999999999995, "socrates": 0.1}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01661376493972915, "einstein": 0.04242473500882781, "hypatia": 0.148076875060247, "ibn_sina": 0.00784778687262939, "sagan": 0.7519964212720947, "socrates": 0.033040416846471944}, {"ibn_sina": 0.01456111550608725, "hypatia": 0.0421339847710306, "sagan": 0.5816417185263132, "socrates": 0.05410110735233753, "darwin": 0.03281398033359728, "einstein": 0.2747480935106341}, {"hypatia": 0.02658232918991473, "darwin": 0.010409789680945525, "sagan": 0.6050110237040526, "ibn_sina": 0.02658232918991473, "einstein": 0.1964183216245217, "socrates": 0.13499620661065073}, {"ibn_sina": 0.01585444848767111, "einstein": 0.027825423651824343, "darwin": 0.031530335766993836, "sagan": 0.5588885741620572, "hypatia": 0.29915149658892437, "socrates": 0.06674972134252916}, {"einstein": 0.042649463252191794, "sagan": 0.588757679460824, "socrates": 0.04832817330835159, "ibn_sina": 0.008939813153541695, "darwin": 0.03321543537838208, "hypatia": 0.27810943544670874}, {"socrates": 0.041865507270677144, "ibn_sina": 0.009341457342001444, "hypatia": 0.27299740029295083, "einstein": 0.06091395364747113, "sagan": 0.577935500955394, "darwin": 0.036946180491505436}], "prompt_tokens": 329, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01661376493972915, "einstein": 0.04242473500882781, "hypatia": 0.148076875060247, "ibn_sina": 0.00784778687262939, "sagan": 0.7519964212720947, "socrates": 0.033040416846471944}, "confidence": 0.5255967261252359, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 171.2512769991008, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "5e16fbf7126da6744631cdedb3c9e498a2b9052c185376919d00b1f279b12f9c", "compiled_ids_sha256": "07c699060a078fa6ae02cca28988321df21cdd9ed54996d09648ee655e350dce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 16.5, 17.75, 14.8125, 19.375, 16.25], "runtime_probabilities": [0.016613764688372612, 0.04242473840713501, 0.1480768620967865, 0.007847786881029606, 0.7519963979721069, 0.03304041549563408], "decide_seconds": 0.16718568699798197, "probabilities_f64_minus_runtime_max_abs": 2.329998771699593e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01456111550608725, "hypatia": 0.0421339847710306, "sagan": 0.5816417185263132, "socrates": 0.05410110735233753, "darwin": 0.03281398033359728, "einstein": 0.2747480935106341}, "confidence": 0.3664954134637193, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 172.03526999946916, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "8a9783359ca22b2ca060f289690fe6ed73f55643bd619baf4637a1b065c7ecca", "compiled_ids_sha256": "2c872e1638c278cbcd624b5ddc049e1eb933fbc9162b13b45a35427b4454fd40", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 16.375, 19.0, 16.625, 16.125, 18.25], "runtime_probabilities": [0.014561115764081478, 0.042133983224630356, 0.581641674041748, 0.05410110205411911, 0.03281398117542267, 0.2747480869293213], "decide_seconds": 0.16810867999811308, "probabilities_f64_minus_runtime_max_abs": 4.448456514261068e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.02658232918991473, "darwin": 0.010409789680945525, "sagan": 0.6050110237040526, "ibn_sina": 0.02658232918991473, "einstein": 0.1964183216245217, "socrates": 0.13499620661065073}, "confidence": 0.36687771469293873, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 171.1363270005677, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "9878e42d49f4c5957122bb2b720a9531598548b6d8d8120ede38748a5a1a24de", "compiled_ids_sha256": "edd99e6f4b88a96477c8055720e7d0ec32a51228358aced1e1cb659383548413", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 15.1875, 19.25, 16.125, 18.125, 17.75], "runtime_probabilities": [0.026582330465316772, 0.01040978915989399, 0.6050110459327698, 0.026582330465316772, 0.19641831517219543, 0.1349962055683136], "decide_seconds": 0.16701212899715756, "probabilities_f64_minus_runtime_max_abs": 2.2228717200256654e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01585444848767111, "einstein": 0.027825423651824343, "darwin": 0.031530335766993836, "sagan": 0.5588885741620572, "hypatia": 0.29915149658892437, "socrates": 0.06674972134252916}, "confidence": 0.36306994426557393, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 175.0572069977352, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "eb5b9290878f211609814ff1f2be0464ce38ec64a9a7a6beac89e8e9aa525329", "compiled_ids_sha256": "681e7bdff5abe93a261513b6c3c8b6333c4c25ec3dfaac932bec6572fc7d713b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 16.0, 16.125, 19.0, 18.375, 16.875], "runtime_probabilities": [0.015854449942708015, 0.02782542258501053, 0.031530339270830154, 0.5588885545730591, 0.2991515100002289, 0.06674972176551819], "decide_seconds": 0.17114094700082205, "probabilities_f64_minus_runtime_max_abs": 1.958899809029191e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.042649463252191794, "sagan": 0.588757679460824, "socrates": 0.04832817330835159, "ibn_sina": 0.008939813153541695, "darwin": 0.03321543537838208, "hypatia": 0.27810943544670874}, "confidence": 0.383830361042682, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 170.94413800077746, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "4440f5d875c123e4ff6fb95694a3ffa472d0c6be75bddc484d62adea6dfb75bb", "compiled_ids_sha256": "2cf7750a23fd10fcd6564f42db1e9309b8cbe4df9456aef8d74e608547e1e0aa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 19.0, 16.5, 14.8125, 16.125, 18.25], "runtime_probabilities": [0.042649466544389725, 0.5887576937675476, 0.04832817614078522, 0.008939812891185284, 0.03321543708443642, 0.27810946106910706], "decide_seconds": 0.16699459900337388, "probabilities_f64_minus_runtime_max_abs": 2.5622398314162353e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.041865507270677144, "ibn_sina": 0.009341457342001444, "hypatia": 0.27299740029295083, "einstein": 0.06091395364747113, "sagan": 0.577935500955394, "darwin": 0.036946180491505436}, "confidence": 0.3636813776853586, "choice": "sagan"}}, "usage": {"input_tokens": 329, "output_tokens": 1}, "latency_ms": 177.9902039997978, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 329, "compiled_prompt_sha256": "977d0b133c97885600b1c793fd8b912d7fdad5da8c8e2667e501ef1416aeaa88", "compiled_ids_sha256": "5135af406a15d382926a370f50fa4a3b44465ddb59226ca7a7f46aa4af9d3568", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 15.25, 18.625, 17.125, 19.375, 16.625], "runtime_probabilities": [0.041865505278110504, 0.009341456927359104, 0.2729974091053009, 0.06091395020484924, 0.5779354572296143, 0.03694617748260498], "decide_seconds": 0.17403438499968615, "probabilities_f64_minus_runtime_max_abs": 4.372577977473924e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:43Z"} {"id": "c-hypatia-01-main-03", "uid": "c048", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "prompt_sha256_by_order": ["518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec"], "request_sha256": "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "compiled_prompt_sha256_by_order": ["1758a9f1601e86873bbd3a7f9387586fd8779e1252ba78aa05c488753ae18af9", "6a674a8716b378f53580b41381308be25f2606421b204bd7b8ffcc49badc30c9", "fc7fa9a6c7732de9f42f45ac6c843b49cf098778d44da80166146a77f93011ac", "1b54e6c415f7d855064ed9713ec139f4b68aca281ae2acdec906924e108ca2da", "0268fe9666c5c235d576b3d8accff25eab943327704299b9aeaa2e91014df2ec", "3e2a04c18bb16bfe4025d0adaece9bd3dc34c5b34f81f8b1db278206a2cedebe"], "compiled_tokens_by_order": [407, 407, 407, 407, 407, 407], "choice": "hypatia", "probs": {"darwin": 0.022117038930681978, "einstein": 0.0038433650724316233, "hypatia": 0.9404404259701094, "ibn_sina": 0.005253257202781719, "sagan": 0.011121145585559553, "socrates": 0.01722476723843587}, "confidence": 0.8264360765827008, "seconds": 0.18062893749993236, "seconds_all_orders": 1.0837736249995942, "seconds_by_order": [0.17737068899805308, 0.18286461699972278, 0.17740507900089142, 0.1845927340000344, 0.1802725070010638, 0.1812679989998287], "argmax_stable": true, "spread": {"darwin": 0.21588224295646283, "einstein": 0.011421234822697513, "hypatia": 0.2832766483765652, "ibn_sina": 0.02442564689739594, "sagan": 0.052900937569319045, "socrates": 0.027087529854528437}, "spread_2dp": {"darwin": 0.21, "einstein": 0.01, "hypatia": 0.28, "ibn_sina": 0.019999999999999997, "sagan": 0.06, "socrates": 0.019999999999999997}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.022117038930681978, "einstein": 0.0038433650724316233, "hypatia": 0.9404404259701094, "ibn_sina": 0.005253257202781719, "sagan": 0.011121145585559553, "socrates": 0.01722476723843587}, {"ibn_sina": 0.02967890410017766, "hypatia": 0.8673447157396492, "sagan": 0.005490045228165497, "socrates": 0.033630604269835, "darwin": 0.04893224048103216, "einstein": 0.014923490181140564}, {"hypatia": 0.9710783442188919, "darwin": 0.0074142746520057656, "sagan": 0.0044969848959717125, "ibn_sina": 0.006965066459381053, "einstein": 0.0035022553584430497, "socrates": 0.006543074415306566}, {"ibn_sina": 0.023833232978311393, "einstein": 0.005660889018701407, "darwin": 0.030602476905992305, "sagan": 0.027006591080954826, "hypatia": 0.8943354695094078, "socrates": 0.018561340506632145}, {"einstein": 0.01406925080738016, "sagan": 0.0035572636874136028, "socrates": 0.008016417466876886, "ibn_sina": 0.007074463584343901, "darwin": 0.040710727073770506, "hypatia": 0.926571877380215}, {"socrates": 0.014274868785615312, "ibn_sina": 0.006334434625377167, "hypatia": 0.6878016958423268, "einstein": 0.011834281881479586, "sagan": 0.05645820125673265, "darwin": 0.22329651760846858}], "prompt_tokens": 407, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.022117038930681978, "einstein": 0.0038433650724316233, "hypatia": 0.9404404259701094, "ibn_sina": 0.005253257202781719, "sagan": 0.011121145585559553, "socrates": 0.01722476723843587}, "confidence": 0.8264360765827008, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.338369001518, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "1758a9f1601e86873bbd3a7f9387586fd8779e1252ba78aa05c488753ae18af9", "compiled_ids_sha256": "c2e4ce3277e1b394388c8fbe6816aea6f422bb1f7b5f4acc516099b73569fad1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 14.625, 20.125, 14.9375, 15.6875, 16.125], "runtime_probabilities": [0.022117041051387787, 0.0038433654699474573, 0.9404405355453491, 0.005253257695585489, 0.011121146380901337, 0.017224770039319992], "decide_seconds": 0.1725160070018319, "probabilities_f64_minus_runtime_max_abs": 1.0957523977062067e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02967890410017766, "hypatia": 0.8673447157396492, "sagan": 0.005490045228165497, "socrates": 0.033630604269835, "darwin": 0.04893224048103216, "einstein": 0.014923490181140564}, "confidence": 0.6758020344717999, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 182.83673699988867, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "6a674a8716b378f53580b41381308be25f2606421b204bd7b8ffcc49badc30c9", "compiled_ids_sha256": "1833fa669c887af0fa128aaaf50de0b1c67feaff6b6de7bac2408f82e6b84ee0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 19.625, 14.5625, 16.375, 16.75, 15.5625], "runtime_probabilities": [0.02967890165746212, 0.8673446774482727, 0.0054900445975363255, 0.033630602061748505, 0.048932239413261414, 0.014923488721251488], "decide_seconds": 0.17816946300081327, "probabilities_f64_minus_runtime_max_abs": 3.829137651401027e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.9710783442188919, "darwin": 0.0074142746520057656, "sagan": 0.0044969848959717125, "ibn_sina": 0.006965066459381053, "einstein": 0.0035022553584430497, "socrates": 0.006543074415306566}, "confidence": 0.9015104483748381, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 177.37698899873067, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "fc7fa9a6c7732de9f42f45ac6c843b49cf098778d44da80166146a77f93011ac", "compiled_ids_sha256": "ba628d8ae8ec01a5e86b8ee55896704e2e2f7edda6088438ce8ad7aef2bc8ecc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 14.875, 14.375, 14.8125, 14.125, 14.75], "runtime_probabilities": [0.9710783362388611, 0.007414274383336306, 0.004496985115110874, 0.006965066306293011, 0.0035022555384784937, 0.006543074734508991], "decide_seconds": 0.172558136000589, "probabilities_f64_minus_runtime_max_abs": 7.98003085833443e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.023833232978311393, "einstein": 0.005660889018701407, "darwin": 0.030602476905992305, "sagan": 0.027006591080954826, "hypatia": 0.8943354695094078, "socrates": 0.018561340506632145}, "confidence": 0.7229205815883974, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 184.56508399685845, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "1b54e6c415f7d855064ed9713ec139f4b68aca281ae2acdec906924e108ca2da", "compiled_ids_sha256": "6cca4592ede3a8d5234c949aecc06cece05232538cfb1cc1a5f5adcce32c90ab", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 14.6875, 16.375, 16.25, 19.75, 15.875], "runtime_probabilities": [0.02383323386311531, 0.005660889204591513, 0.030602479353547096, 0.027006594464182854, 0.8943355679512024, 0.018561340868473053], "decide_seconds": 0.17990495000049123, "probabilities_f64_minus_runtime_max_abs": 9.844179460483105e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.01406925080738016, "sagan": 0.0035572636874136028, "socrates": 0.008016417466876886, "ibn_sina": 0.007074463584343901, "darwin": 0.040710727073770506, "hypatia": 0.926571877380215}, "confidence": 0.8020086150227799, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 180.2441869986069, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "0268fe9666c5c235d576b3d8accff25eab943327704299b9aeaa2e91014df2ec", "compiled_ids_sha256": "dc81e9218eab3129e8be26b6f35a94acc39ea2c268947a6304cdda3dbc9f04d6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.1875, 15.0, 14.875, 16.625, 19.75], "runtime_probabilities": [0.014069249853491783, 0.0035572636406868696, 0.008016416803002357, 0.007074463181197643, 0.040710724890232086, 0.9265718460083008], "decide_seconds": 0.17556263300139108, "probabilities_f64_minus_runtime_max_abs": 3.137191417756924e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.014274868785615312, "ibn_sina": 0.006334434625377167, "hypatia": 0.6878016958423268, "einstein": 0.011834281881479586, "sagan": 0.05645820125673265, "darwin": 0.22329651760846858}, "confidence": 0.49787158082308747, "choice": "hypatia"}}, "usage": {"input_tokens": 407, "output_tokens": 1}, "latency_ms": 181.2407599973085, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 407, "compiled_prompt_sha256": "3e2a04c18bb16bfe4025d0adaece9bd3dc34c5b34f81f8b1db278206a2cedebe", "compiled_ids_sha256": "c3d418fa8f304a9c23bf22affdb273fe008afd503efd7edd5ba1462d0fe2b0e7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.875, 15.0625, 19.75, 15.6875, 17.25, 18.625], "runtime_probabilities": [0.014274868182837963, 0.006334434263408184, 0.6878016591072083, 0.011834281496703625, 0.05645819753408432, 0.22329650819301605], "decide_seconds": 0.17652532599822734, "probabilities_f64_minus_runtime_max_abs": 3.673511850266209e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:44Z"} {"id": "c-hypatia-01-starter-01", "uid": "c049", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "prompt_sha256_by_order": ["2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e"], "request_sha256": "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "compiled_prompt_sha256_by_order": ["86edd11982c21ff733510b65fb8061cf99f203a5783f0d267f8b7922baef66d7", "9ca6df4d96fbece59b6a36b86ab04b90e3c8ba3e79e2b24eb8a990f20c891ef4", "41d3926c51b47a1f4a3e6fb517b1394348ef7a9ff3fe5a47cc6fe56ec59913b5", "090512e592eb25c1371a09a8a7c86aa8936430c30e71b7b28dcd17c102105fbc", "2c5d1055ff0406775e9fa226204eb617ecd53903ae204b80e077dd04042fe029", "6e859a1ee6e11bc8c3827dfa923f46015a885917a8f0c733a1a062066aca9189"], "compiled_tokens_by_order": [272, 272, 272, 272, 272, 272], "choice": "hypatia", "probs": {"darwin": 0.010733826728078865, "einstein": 0.005397307031346558, "hypatia": 0.33391956188101823, "ibn_sina": 0.33391956188101823, "sagan": 0.021346763406134336, "socrates": 0.29468297907240376}, "confidence": 0.30149355322003546, "seconds": 0.16591499883361394, "seconds_all_orders": 0.9954899930016836, "seconds_by_order": [0.16480016599962255, 0.16475149600228178, 0.16549535000012838, 0.16840770799899474, 0.16745204600010766, 0.1645832270005485], "argmax_stable": false, "spread": {"darwin": 0.03644225142166656, "einstein": 0.00483431028095101, "hypatia": 0.5267677498935766, "ibn_sina": 0.29579031780429293, "sagan": 0.0211292412278952, "socrates": 0.5449251660311801}, "spread_2dp": {"darwin": 0.04, "einstein": 0.0, "hypatia": 0.53, "ibn_sina": 0.29000000000000004, "sagan": 0.02, "socrates": 0.5399999999999999}, "choices_by_order": ["hypatia", "hypatia", "socrates", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.010733826728078865, "einstein": 0.005397307031346558, "hypatia": 0.33391956188101823, "ibn_sina": 0.33391956188101823, "sagan": 0.021346763406134336, "socrates": 0.29468297907240376}, {"ibn_sina": 0.03812924407672528, "hypatia": 0.7658463397563002, "sagan": 0.007508093325417567, "socrates": 0.17088341643891206, "darwin": 0.007992323741945164, "einstein": 0.009640582660699743}, {"hypatia": 0.23907858986272365, "darwin": 0.0076851686781210585, "sagan": 0.008180819473708958, "ibn_sina": 0.16431615160111396, "einstein": 0.007219547846151895, "socrates": 0.5735197225381804}, {"ibn_sina": 0.055557613840833554, "einstein": 0.00550094870694222, "darwin": 0.004024580048032749, "sagan": 0.004560446655671922, "hypatia": 0.4651782053742598, "socrates": 0.4651782053742598}, {"einstein": 0.008720847013569769, "sagan": 0.0034151327599014803, "socrates": 0.028594556507000237, "ibn_sina": 0.21128678215426722, "darwin": 0.010519349468340478, "hypatia": 0.7374633320969208}, {"socrates": 0.05887889236521415, "ibn_sina": 0.23287053666500457, "hypatia": 0.6330077481999877, "einstein": 0.010231617312297568, "sagan": 0.02454437398779668, "darwin": 0.04046683146969931}], "prompt_tokens": 272, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.010733826728078865, "einstein": 0.005397307031346558, "hypatia": 0.33391956188101823, "ibn_sina": 0.33391956188101823, "sagan": 0.021346763406134336, "socrates": 0.29468297907240376}, "confidence": 0.30149355322003546, "choice": "hypatia"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 164.7734660000424, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "86edd11982c21ff733510b65fb8061cf99f203a5783f0d267f8b7922baef66d7", "compiled_ids_sha256": "0632451188e0cd8274e38703ba69b1d81b731677e7550055093537363c0d1780", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.75, 18.875, 18.875, 16.125, 18.75], "runtime_probabilities": [0.010733827017247677, 0.005397306755185127, 0.33391955494880676, 0.33391955494880676, 0.021346760913729668, 0.29468294978141785], "decide_seconds": 0.1612337740007206, "probabilities_f64_minus_runtime_max_abs": 2.929098591453183e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03812924407672528, "hypatia": 0.7658463397563002, "sagan": 0.007508093325417567, "socrates": 0.17088341643891206, "darwin": 0.007992323741945164, "einstein": 0.009640582660699743}, "confidence": 0.5809403785318339, "choice": "hypatia"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 164.72925600101007, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "9ca6df4d96fbece59b6a36b86ab04b90e3c8ba3e79e2b24eb8a990f20c891ef4", "compiled_ids_sha256": "466504b359a0b9c3df1f7bf4cb4364e60600e74cc215d009d40ebeb38e0a21f3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 19.75, 15.125, 18.25, 15.1875, 15.375], "runtime_probabilities": [0.03812924399971962, 0.7658463716506958, 0.00750809395685792, 0.1708834320306778, 0.00799232441931963, 0.009640583768486977], "decide_seconds": 0.16141838199837366, "probabilities_f64_minus_runtime_max_abs": 3.189439556905427e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.23907858986272365, "darwin": 0.0076851686781210585, "sagan": 0.008180819473708958, "ibn_sina": 0.16431615160111396, "einstein": 0.007219547846151895, "socrates": 0.5735197225381804}, "confidence": 0.40279472677656947, "choice": "socrates"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 165.47254000033718, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "41d3926c51b47a1f4a3e6fb517b1394348ef7a9ff3fe5a47cc6fe56ec59913b5", "compiled_ids_sha256": "70b1a9174b9a981f2b65e1fa234015055ab28c6563e08586231b72dd78c54f11", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.75, 15.3125, 15.375, 18.375, 15.25, 19.625], "runtime_probabilities": [0.2390785813331604, 0.007685168646275997, 0.00818081945180893, 0.16431616246700287, 0.007219547871500254, 0.5735197067260742], "decide_seconds": 0.1619295280033839, "probabilities_f64_minus_runtime_max_abs": 1.5812106179957652e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.055557613840833554, "einstein": 0.00550094870694222, "darwin": 0.004024580048032749, "sagan": 0.004560446655671922, "hypatia": 0.4651782053742598, "socrates": 0.4651782053742598}, "confidence": 0.47090335726515675, "choice": "hypatia"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 168.38547799852677, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "090512e592eb25c1371a09a8a7c86aa8936430c30e71b7b28dcd17c102105fbc", "compiled_ids_sha256": "def3bd76ed0253733af5bf1344c531965c9d65e5944e97016387231d2cc5138b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 15.0625, 14.75, 14.875, 19.5, 19.5], "runtime_probabilities": [0.05555761605501175, 0.0055009485222399235, 0.004024580121040344, 0.004560446832329035, 0.4651782214641571, 0.4651782214641571], "decide_seconds": 0.16500332400028128, "probabilities_f64_minus_runtime_max_abs": 1.6089897303395162e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.008720847013569769, "sagan": 0.0034151327599014803, "socrates": 0.028594556507000237, "ibn_sina": 0.21128678215426722, "darwin": 0.010519349468340478, "hypatia": 0.7374633320969208}, "confidence": 0.5739707013732668, "choice": "hypatia"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 167.42969599727076, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "2c5d1055ff0406775e9fa226204eb617ecd53903ae204b80e077dd04042fe029", "compiled_ids_sha256": "8d923a5efcfabbc03486d7d11fba9defcd41cde57b8767d84921dfb0ee5276ae", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.5, 16.625, 18.625, 15.625, 19.875], "runtime_probabilities": [0.00872084591537714, 0.003415132872760296, 0.028594553470611572, 0.2112867683172226, 0.010519348084926605, 0.7374632954597473], "decide_seconds": 0.16406685099718743, "probabilities_f64_minus_runtime_max_abs": 3.6637173517206634e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.05887889236521415, "ibn_sina": 0.23287053666500457, "hypatia": 0.6330077481999877, "einstein": 0.010231617312297568, "sagan": 0.02454437398779668, "darwin": 0.04046683146969931}, "confidence": 0.4065950457199148, "choice": "hypatia"}}, "usage": {"input_tokens": 272, "output_tokens": 1}, "latency_ms": 164.56213800120167, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 272, "compiled_prompt_sha256": "6e859a1ee6e11bc8c3827dfa923f46015a885917a8f0c733a1a062066aca9189", "compiled_ids_sha256": "7a99cb6ba0a11ada1276eb69a6695f8482f07b8bf729e14789d4bb0ebd66785f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 18.5, 19.5, 15.375, 16.25, 16.75], "runtime_probabilities": [0.05887888744473457, 0.23287054896354675, 0.6330077052116394, 0.010231616906821728, 0.02454437129199505, 0.040466826409101486], "decide_seconds": 0.16118916300183628, "probabilities_f64_minus_runtime_max_abs": 4.2988348325501136e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:45Z"} {"id": "c-hypatia-01-starter-01", "uid": "c050", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "prompt_sha256_by_order": ["49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7"], "request_sha256": "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "compiled_prompt_sha256_by_order": ["ecbc5708033f479cd2caa368a60bec322a499a7331fb8a6eacdc2dac8b96a819", "8f0580154b4096fd98b33a416ca21c762667e74d33ce54f82dd54523370fe337", "d2bb427dc7e60d790490212202c7e2fbe8521778e243040981565edde2647c70", "a3b372a62776ed9ecaba5285e18e91a63a3492e8a908a6b398380a222cfd691e", "1ff82cbd9a8fa813d552587570252c629dc21bc8e01a98dacff34dfa6fc901b8", "c87a85d26c8641910f485c0600d3d5a19d21cbf7c411c8c88c492ba178ddad9a"], "compiled_tokens_by_order": [326, 326, 326, 326, 326, 326], "choice": "ibn_sina", "probs": {"darwin": 0.0062366888576695055, "einstein": 0.003782743007264715, "hypatia": 0.008008067009220882, "ibn_sina": 0.9256066957065711, "sagan": 0.046083243840895696, "socrates": 0.010282561578378199}, "confidence": 0.8036265564604589, "seconds": 0.1721817891672496, "seconds_all_orders": 1.0330907350034977, "seconds_by_order": [0.17110586800117744, 0.17071159000261105, 0.17462718000024324, 0.17066329099907307, 0.17525657499936642, 0.17072623100102646], "argmax_stable": false, "spread": {"darwin": 0.12352108157266362, "einstein": 0.024629041986869056, "hypatia": 0.033330836231222, "ibn_sina": 0.5794802966027308, "sagan": 0.495949546576315, "socrates": 0.15321597237896348}, "spread_2dp": {"darwin": 0.12000000000000001, "einstein": 0.03, "hypatia": 0.03, "ibn_sina": 0.5800000000000001, "sagan": 0.5, "socrates": 0.15}, "choices_by_order": ["ibn_sina", "sagan", "ibn_sina", "sagan", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0062366888576695055, "einstein": 0.003782743007264715, "hypatia": 0.008008067009220882, "ibn_sina": 0.9256066957065711, "sagan": 0.046083243840895696, "socrates": 0.010282561578378199}, {"ibn_sina": 0.409627049798666, "hypatia": 0.01690732755495251, "sagan": 0.5259715433043023, "socrates": 0.02310957678596229, "darwin": 0.015882964362388615, "einstein": 0.008501538193728344}, {"hypatia": 0.016790536101319885, "darwin": 0.007450755229779961, "sagan": 0.1165494213186378, "ibn_sina": 0.7599976949979834, "einstein": 0.008442811762804729, "socrates": 0.0907687805894742}, {"ibn_sina": 0.3461263991038403, "einstein": 0.02841178499413377, "darwin": 0.02841178499413377, "sagan": 0.3922125937101075, "hypatia": 0.041338903240442884, "socrates": 0.16349853395734168}, {"einstein": 0.009156191684335922, "sagan": 0.030021996727987273, "socrates": 0.04368174748891722, "ibn_sina": 0.877371352058007, "darwin": 0.030021996727987273, "hypatia": 0.009746715312765372}, {"socrates": 0.0175608046133405, "ibn_sina": 0.7467044129509323, "hypatia": 0.02254851946101655, "einstein": 0.004726426442427204, "sagan": 0.07870206610195038, "darwin": 0.12975777043033312}], "prompt_tokens": 326, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0062366888576695055, "einstein": 0.003782743007264715, "hypatia": 0.008008067009220882, "ibn_sina": 0.9256066957065711, "sagan": 0.046083243840895696, "socrates": 0.010282561578378199}, "confidence": 0.8036265564604589, "choice": "ibn_sina"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 171.06448800041107, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "ecbc5708033f479cd2caa368a60bec322a499a7331fb8a6eacdc2dac8b96a819", "compiled_ids_sha256": "3213fbf2a47076131e60a1856d79082d20d04b59cb63f46819df660c2b4096ff", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.25, 15.0, 19.75, 16.75, 15.25], "runtime_probabilities": [0.006236689165234566, 0.0037827431224286556, 0.00800806749612093, 0.9256067276000977, 0.0460832417011261, 0.010282562114298344], "decide_seconds": 0.16716479699971387, "probabilities_f64_minus_runtime_max_abs": 3.1893526597492894e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.409627049798666, "hypatia": 0.01690732755495251, "sagan": 0.5259715433043023, "socrates": 0.02310957678596229, "darwin": 0.015882964362388615, "einstein": 0.008501538193728344}, "confidence": 0.4609150763972739, "choice": "sagan"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 170.68804100199486, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "8f0580154b4096fd98b33a416ca21c762667e74d33ce54f82dd54523370fe337", "compiled_ids_sha256": "3c5d83ceabce936b6414b34a3cdca48c14d190eac44812367c72567d2a62cd9b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.875, 15.6875, 19.125, 16.0, 15.625, 15.0], "runtime_probabilities": [0.4096270799636841, 0.016907328739762306, 0.5259715914726257, 0.02310957759618759, 0.015882965177297592, 0.008501538075506687], "decide_seconds": 0.16682209099963075, "probabilities_f64_minus_runtime_max_abs": 4.8168323441011296e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.016790536101319885, "darwin": 0.007450755229779961, "sagan": 0.1165494213186378, "ibn_sina": 0.7599976949979834, "einstein": 0.008442811762804729, "socrates": 0.0907687805894742}, "confidence": 0.5410544600387993, "choice": "ibn_sina"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 174.60398999901372, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "d2bb427dc7e60d790490212202c7e2fbe8521778e243040981565edde2647c70", "compiled_ids_sha256": "18e44558ab61bf0094ab3491ca6525c0e8ceff8031aac6cfedaf27a0b4fc2ba2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 15.125, 17.875, 19.75, 15.25, 17.625], "runtime_probabilities": [0.016790537163615227, 0.007450756151229143, 0.11654943227767944, 0.7599977254867554, 0.008442812599241734, 0.09076878428459167], "decide_seconds": 0.1707950800009712, "probabilities_f64_minus_runtime_max_abs": 3.048877195155342e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.3461263991038403, "einstein": 0.02841178499413377, "darwin": 0.02841178499413377, "sagan": 0.3922125937101075, "hypatia": 0.041338903240442884, "socrates": 0.16349853395734168}, "confidence": 0.23848446214407681, "choice": "sagan"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 170.6402510026237, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "a3b372a62776ed9ecaba5285e18e91a63a3492e8a908a6b398380a222cfd691e", "compiled_ids_sha256": "fe1689c093e36ce7127e581f5f58900e0f5411ab3498d73285e79b5f1e8648cf", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 16.0, 16.0, 18.625, 16.375, 17.75], "runtime_probabilities": [0.34612637758255005, 0.028411785140633583, 0.028411785140633583, 0.39221256971359253, 0.04133889824151993, 0.16349852085113525], "decide_seconds": 0.1668218300001172, "probabilities_f64_minus_runtime_max_abs": 2.399651499729316e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.009156191684335922, "sagan": 0.030021996727987273, "socrates": 0.04368174748891722, "ibn_sina": 0.877371352058007, "darwin": 0.030021996727987273, "hypatia": 0.009746715312765372}, "confidence": 0.6929532032465042, "choice": "ibn_sina"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 175.234205002198, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "1ff82cbd9a8fa813d552587570252c629dc21bc8e01a98dacff34dfa6fc901b8", "compiled_ids_sha256": "ef41420cca1207458ea2379a7b8678012efd36f35122fbf25370351e2cbee376", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 16.375, 16.75, 19.75, 16.375, 15.25], "runtime_probabilities": [0.00915619172155857, 0.030021995306015015, 0.04368174448609352, 0.8773713707923889, 0.030021995306015015, 0.009746716357767582], "decide_seconds": 0.17147741499866243, "probabilities_f64_minus_runtime_max_abs": 1.8734381934670807e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0175608046133405, "ibn_sina": 0.7467044129509323, "hypatia": 0.02254851946101655, "einstein": 0.004726426442427204, "sagan": 0.07870206610195038, "darwin": 0.12975777043033312}, "confidence": 0.5172666855484083, "choice": "ibn_sina"}}, "usage": {"input_tokens": 326, "output_tokens": 1}, "latency_ms": 170.70280100233504, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 326, "compiled_prompt_sha256": "c87a85d26c8641910f485c0600d3d5a19d21cbf7c411c8c88c492ba178ddad9a", "compiled_ids_sha256": "ce26786a3b5b708763889c8e40308deb955fcfd36bbcd5adc0a035bc38d423c1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 19.375, 15.875, 14.3125, 17.125, 17.625], "runtime_probabilities": [0.01756080612540245, 0.7467044591903687, 0.02254852093756199, 0.004726426675915718, 0.07870206981897354, 0.12975777685642242], "decide_seconds": 0.166853610000544, "probabilities_f64_minus_runtime_max_abs": 4.623943639092687e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:46Z"} {"id": "c-hypatia-01-starter-02", "uid": "c051", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "prompt_sha256_by_order": ["d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd"], "request_sha256": "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "compiled_prompt_sha256_by_order": ["d3cc18c537cd2524953ff31d3b343b481a694d8832a6516cf95f0af6ac3a93fc", "e5e94315a854703a398e48adb212aee85dbcddd9bfa474bab8d009899505f94f", "f5114197072d0456f7f044d5682ca0a5e97b2cc3e89232c198233ea48eea91a3", "316e4e233f9cec8ca26a21467e737810ce71f6e02d434c3980ecd0a7a99e5cdb", "2b8889e768b472422cf59693e3e301d0243bcba7fc27cbe5e6a0f9fbb85d85ce", "b0c5bf7d8ff8c34974293c92ef2fbd6781c4f2e077caba91ae21a56bb5a9ee7b"], "compiled_tokens_by_order": [303, 303, 303, 303, 303, 303], "choice": "hypatia", "probs": {"darwin": 0.014398575931199371, "einstein": 0.004391317552993604, "hypatia": 0.6937622399144565, "ibn_sina": 0.1366098848549696, "sagan": 0.09389050923768982, "socrates": 0.05694747250869111}, "confidence": 0.44423775761147666, "seconds": 0.16816667483241568, "seconds_all_orders": 1.0090000489944941, "seconds_by_order": [0.16542794099950697, 0.17031406299793161, 0.16970613800003775, 0.16545876099917223, 0.17049102199962363, 0.16760212399822194], "argmax_stable": true, "spread": {"darwin": 0.0599963697403586, "einstein": 0.01443198903855038, "hypatia": 0.4338389863177389, "ibn_sina": 0.14174069213512447, "sagan": 0.09099463009375447, "socrates": 0.16850924976037468}, "spread_2dp": {"darwin": 0.06, "einstein": 0.02, "hypatia": 0.43999999999999995, "ibn_sina": 0.13999999999999999, "sagan": 0.09, "socrates": 0.16}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.014398575931199371, "einstein": 0.004391317552993604, "hypatia": 0.6937622399144565, "ibn_sina": 0.1366098848549696, "sagan": 0.09389050923768982, "socrates": 0.05694747250869111}, {"ibn_sina": 0.15319411632969437, "hypatia": 0.5347000050559388, "sagan": 0.05635696590612089, "socrates": 0.17359167593793262, "darwin": 0.06386080873606076, "einstein": 0.01829642803425254}, {"hypatia": 0.9532945634811288, "darwin": 0.00500243076170681, "sagan": 0.00500243076170681, "ibn_sina": 0.016402339000489936, "einstein": 0.0038958969944777474, "socrates": 0.016402339000489936}, {"ibn_sina": 0.08262534896323197, "einstein": 0.0038644389957021596, "darwin": 0.0038644389957021596, "sagan": 0.019625240563794923, "hypatia": 0.7839274843500847, "socrates": 0.10609304813148399}, {"einstein": 0.005082426177557955, "sagan": 0.002895879143935359, "socrates": 0.005082426177557955, "ibn_sina": 0.011453424194569906, "darwin": 0.006946852932701122, "hypatia": 0.9685389913736777}, {"socrates": 0.018205664950541964, "ibn_sina": 0.056077395926415814, "hypatia": 0.8771980615007959, "einstein": 0.0043242244512302005, "sagan": 0.03001606705119833, "darwin": 0.014178586119817711}], "prompt_tokens": 303, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.014398575931199371, "einstein": 0.004391317552993604, "hypatia": 0.6937622399144565, "ibn_sina": 0.1366098848549696, "sagan": 0.09389050923768982, "socrates": 0.05694747250869111}, "confidence": 0.44423775761147666, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 165.40169099971536, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "d3cc18c537cd2524953ff31d3b343b481a694d8832a6516cf95f0af6ac3a93fc", "compiled_ids_sha256": "4176637f3b7f7e0c1eaf039089f3e571cafaabcf2dff27018fe9ae1943bf4ef4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 14.5625, 19.625, 18.0, 17.625, 17.125], "runtime_probabilities": [0.014398575760424137, 0.004391317255795002, 0.6937622427940369, 0.13660988211631775, 0.09389051049947739, 0.05694747343659401], "decide_seconds": 0.1617017400021723, "probabilities_f64_minus_runtime_max_abs": 2.8795803519443552e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.15319411632969437, "hypatia": 0.5347000050559388, "sagan": 0.05635696590612089, "socrates": 0.17359167593793262, "darwin": 0.06386080873606076, "einstein": 0.01829642803425254}, "confidence": 0.2537555582398282, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 170.29031299898634, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "e5e94315a854703a398e48adb212aee85dbcddd9bfa474bab8d009899505f94f", "compiled_ids_sha256": "661f53f3bee1418a76f3203da4a844064802ab82de09e99b9e4e14f4acf44db4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.75, 19.0, 16.75, 17.875, 16.875, 15.625], "runtime_probabilities": [0.1531941294670105, 0.5347000360488892, 0.056356966495513916, 0.17359167337417603, 0.06386081129312515, 0.01829642802476883], "decide_seconds": 0.16665041200030828, "probabilities_f64_minus_runtime_max_abs": 3.099295031727678e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.9532945634811288, "darwin": 0.00500243076170681, "sagan": 0.00500243076170681, "ibn_sina": 0.016402339000489936, "einstein": 0.0038958969944777474, "socrates": 0.016402339000489936}, "confidence": 0.8576520467354621, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 169.68207800164237, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "f5114197072d0456f7f044d5682ca0a5e97b2cc3e89232c198233ea48eea91a3", "compiled_ids_sha256": "07c031aafebb4d50147cbd8c37e2b93c77bfe47f2a5860848d67bcd2dd338a9c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.875, 14.625, 14.625, 15.8125, 14.375, 15.8125], "runtime_probabilities": [0.9532946944236755, 0.005002431571483612, 0.005002431571483612, 0.016402339562773705, 0.0038958974182605743, 0.016402339562773705], "decide_seconds": 0.16580384799817693, "probabilities_f64_minus_runtime_max_abs": 1.3094254669443472e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.08262534896323197, "einstein": 0.0038644389957021596, "darwin": 0.0038644389957021596, "sagan": 0.019625240563794923, "hypatia": 0.7839274843500847, "socrates": 0.10609304813148399}, "confidence": 0.5786490091636819, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 165.4353310004808, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "316e4e233f9cec8ca26a21467e737810ce71f6e02d434c3980ecd0a7a99e5cdb", "compiled_ids_sha256": "5a8cdc19d6b5add526023fa46e680b4cf0175abacfd443699961c58b5f52d7dc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 14.3125, 14.3125, 15.9375, 19.625, 17.625], "runtime_probabilities": [0.08262534439563751, 0.003864438971504569, 0.003864438971504569, 0.019625240936875343, 0.7839275002479553, 0.10609304904937744], "decide_seconds": 0.16169055000136723, "probabilities_f64_minus_runtime_max_abs": 1.5897870575543038e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.005082426177557955, "sagan": 0.002895879143935359, "socrates": 0.005082426177557955, "ibn_sina": 0.011453424194569906, "darwin": 0.006946852932701122, "hypatia": 0.9685389913736777}, "confidence": 0.8954720378378789, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 170.46875300002284, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "2b8889e768b472422cf59693e3e301d0243bcba7fc27cbe5e6a0f9fbb85d85ce", "compiled_ids_sha256": "f80051bc116833e78d71fc8244e831b2d2def084d69f32d6a13fd283ba03f22e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 14.0625, 14.625, 15.4375, 14.9375, 19.875], "runtime_probabilities": [0.005082426592707634, 0.0028958795592188835, 0.005082426592707634, 0.011453424580395222, 0.006946852896362543, 0.9685390591621399], "decide_seconds": 0.16685368999969796, "probabilities_f64_minus_runtime_max_abs": 6.778846217692092e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.018205664950541964, "ibn_sina": 0.056077395926415814, "hypatia": 0.8771980615007959, "einstein": 0.0043242244512302005, "sagan": 0.03001606705119833, "darwin": 0.014178586119817711}, "confidence": 0.6994319781859571, "choice": "hypatia"}}, "usage": {"input_tokens": 303, "output_tokens": 1}, "latency_ms": 167.57996500018635, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 303, "compiled_prompt_sha256": "b0c5bf7d8ff8c34974293c92ef2fbd6781c4f2e077caba91ae21a56bb5a9ee7b", "compiled_ids_sha256": "7efcb9b8a23db282a2a4c1f4e50da3b790d13754decda6c8aa3e8f85867e9ee4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 17.25, 20.0, 14.6875, 16.625, 15.875], "runtime_probabilities": [0.018205665051937103, 0.05607739835977554, 0.8771981596946716, 0.004324224777519703, 0.030016068369150162, 0.01417858712375164], "decide_seconds": 0.1639768120003282, "probabilities_f64_minus_runtime_max_abs": 9.819387569720561e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:47Z"} {"id": "c-hypatia-01-starter-02", "uid": "c052", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "prompt_sha256_by_order": ["44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d"], "request_sha256": "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "compiled_prompt_sha256_by_order": ["25078b32820a0c397cc2958b1cb9a27364e725d270a1e669793c995eab4ad072", "04867182226b2f863c7c6e001eb7a58b3bdf7b4d7644608ae220fb5bf0b7ff68", "ac59f35e73dad07af18bbc91f9c6c3bb5ff9b3bb081a4434c68a9758c0c4dc98", "3939c699cb8a8b20c876788fd495fa2bb581079d01287b27cc86cf7f299af77b", "5b257bb68c107e50dc9e2333fab374b26a415d3d36cb8718177d6c0312b8ca47", "eba7e53d2e352a92a741c6b4cdd60bc167e7ccc62b81d6aa2c2351260e3c586c"], "compiled_tokens_by_order": [273, 273, 273, 273, 273, 273], "choice": "hypatia", "probs": {"darwin": 0.008451436992470543, "einstein": 0.005126055654562914, "hypatia": 0.8098398282336456, "ibn_sina": 0.10959990253029021, "sagan": 0.031400897810337426, "socrates": 0.03558187877869335}, "confidence": 0.6049273141261942, "seconds": 0.16562802816649005, "seconds_all_orders": 0.9937681689989404, "seconds_by_order": [0.16693164999742294, 0.16510411400304292, 0.16703967899957206, 0.16512284399868804, 0.16494370500004152, 0.1646261770001729], "argmax_stable": true, "spread": {"darwin": 0.007361747179299766, "einstein": 0.005288726380483629, "hypatia": 0.18084166238426014, "ibn_sina": 0.10579651136507169, "sagan": 0.030086480995204938, "socrates": 0.12289985016158217}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.17999999999999994, "ibn_sina": 0.11, "sagan": 0.03, "socrates": 0.12}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.008451436992470543, "einstein": 0.005126055654562914, "hypatia": 0.8098398282336456, "ibn_sina": 0.10959990253029021, "sagan": 0.031400897810337426, "socrates": 0.03558187877869335}, {"ibn_sina": 0.003803391165218529, "hypatia": 0.9906814906179058, "sagan": 0.0013144168151324874, "socrates": 0.0017965947734399077, "darwin": 0.0010896898131707764, "einstein": 0.0013144168151324874}, {"hypatia": 0.8648410885755995, "darwin": 0.0054741979806805395, "sagan": 0.01158887721604177, "ibn_sina": 0.09115358010510681, "einstein": 0.006603143195616116, "socrates": 0.02033911292695521}, {"ibn_sina": 0.04048300851896951, "einstein": 0.005832130338678052, "darwin": 0.004266877705830735, "sagan": 0.011598576132016704, "hypatia": 0.813122962369483, "socrates": 0.12469644493502208}, {"einstein": 0.003271830149275379, "sagan": 0.0019844652989076794, "socrates": 0.004472061585105914, "ibn_sina": 0.0227110026892436, "darwin": 0.0018642326244939231, "hypatia": 0.9656964076529736}, {"socrates": 0.008262637915346559, "ibn_sina": 0.019821037958120666, "hypatia": 0.9550310917525453, "einstein": 0.0034443799292635867, "sagan": 0.007762029990974463, "darwin": 0.005678822453749478}], "prompt_tokens": 273, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.008451436992470543, "einstein": 0.005126055654562914, "hypatia": 0.8098398282336456, "ibn_sina": 0.10959990253029021, "sagan": 0.031400897810337426, "socrates": 0.03558187877869335}, "confidence": 0.6049273141261942, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 166.90586000186158, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "25078b32820a0c397cc2958b1cb9a27364e725d270a1e669793c995eab4ad072", "compiled_ids_sha256": "9403e074d597096fdde8ed17e223082b95d8f4cb2f6a7e59d08436d83892a5ca", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 14.6875, 19.75, 17.75, 16.5, 16.625], "runtime_probabilities": [0.008451436646282673, 0.005126055330038071, 0.8098397850990295, 0.10959989577531815, 0.0314008966088295, 0.03558187931776047], "decide_seconds": 0.16339701600008993, "probabilities_f64_minus_runtime_max_abs": 4.313461610117031e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.003803391165218529, "hypatia": 0.9906814906179058, "sagan": 0.0013144168151324874, "socrates": 0.0017965947734399077, "darwin": 0.0010896898131707764, "einstein": 0.0013144168151324874}, "confidence": 0.9627745230456232, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 165.08198399969842, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "04867182226b2f863c7c6e001eb7a58b3bdf7b4d7644608ae220fb5bf0b7ff68", "compiled_ids_sha256": "b2253b514a00fbf14563cf99c55e250e87a609c9d73ef595c32edd686b839250", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 20.25, 13.625, 13.9375, 13.4375, 13.625], "runtime_probabilities": [0.003803391009569168, 0.9906814694404602, 0.0013144167605787516, 0.001796594588086009, 0.001089689671061933, 0.0013144167605787516], "decide_seconds": 0.1616100009996444, "probabilities_f64_minus_runtime_max_abs": 2.117744557494916e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.8648410885755995, "darwin": 0.0054741979806805395, "sagan": 0.01158887721604177, "ibn_sina": 0.09115358010510681, "einstein": 0.006603143195616116, "socrates": 0.02033911292695521}, "confidence": 0.7005975283889232, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 167.01807900244603, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "ac59f35e73dad07af18bbc91f9c6c3bb5ff9b3bb081a4434c68a9758c0c4dc98", "compiled_ids_sha256": "4993b0df859d5873143dcbb41ea55058d7b3c8fee204d3886f6b78a775d9f913", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.875, 14.8125, 15.5625, 17.625, 15.0, 16.125], "runtime_probabilities": [0.8648411631584167, 0.005474198143929243, 0.011588877998292446, 0.091153584420681, 0.00660314317792654, 0.0203391145914793], "decide_seconds": 0.16366410499904305, "probabilities_f64_minus_runtime_max_abs": 7.458281725547522e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04048300851896951, "einstein": 0.005832130338678052, "darwin": 0.004266877705830735, "sagan": 0.011598576132016704, "hypatia": 0.813122962369483, "socrates": 0.12469644493502208}, "confidence": 0.6301852972468045, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 165.1014339986432, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "3939c699cb8a8b20c876788fd495fa2bb581079d01287b27cc86cf7f299af77b", "compiled_ids_sha256": "310ec05141b06cb9f4723f00de0b2705dcadfcc935d5699672e3d80f9a6efa76", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 15.0625, 14.75, 15.75, 20.0, 18.125], "runtime_probabilities": [0.040483009070158005, 0.005832130555063486, 0.0042668781243264675, 0.011598576791584492, 0.8131229877471924, 0.12469645589590073], "decide_seconds": 0.16170988899830263, "probabilities_f64_minus_runtime_max_abs": 2.5377709378382463e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.003271830149275379, "sagan": 0.0019844652989076794, "socrates": 0.004472061585105914, "ibn_sina": 0.0227110026892436, "darwin": 0.0018642326244939231, "hypatia": 0.9656964076529736}, "confidence": 0.8958296394234823, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.92002499944647, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "5b257bb68c107e50dc9e2333fab374b26a415d3d36cb8718177d6c0312b8ca47", "compiled_ids_sha256": "0d8258aa2299d8d5fe7cb95bafb958f54ed427600a110a229e2ae0f5e3e571c1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.8125, 14.625, 16.25, 13.75, 20.0], "runtime_probabilities": [0.003271830268204212, 0.0019844654016196728, 0.004472061991691589, 0.022711003199219704, 0.001864232704974711, 0.9656964540481567], "decide_seconds": 0.1613919720002741, "probabilities_f64_minus_runtime_max_abs": 4.639518313975799e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.008262637915346559, "ibn_sina": 0.019821037958120666, "hypatia": 0.9550310917525453, "einstein": 0.0034443799292635867, "sagan": 0.007762029990974463, "darwin": 0.005678822453749478}, "confidence": 0.8616450421082609, "choice": "hypatia"}}, "usage": {"input_tokens": 273, "output_tokens": 1}, "latency_ms": 164.60347700194689, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 273, "compiled_prompt_sha256": "eba7e53d2e352a92a741c6b4cdd60bc167e7ccc62b81d6aa2c2351260e3c586c", "compiled_ids_sha256": "4ee5763d44578af9946676e2779063aa20d3ae5a4788dd98641c203c5f718886", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 16.25, 20.125, 14.5, 15.3125, 15.0], "runtime_probabilities": [0.008262638002634048, 0.019821036607027054, 0.9550310373306274, 0.0034443798940628767, 0.007762030232697725, 0.005678822286427021], "decide_seconds": 0.16125001399996108, "probabilities_f64_minus_runtime_max_abs": 5.442191786109163e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:48Z"} {"id": "c-hypatia-01-starter-02", "uid": "c053", "task": "kperm", "label": "hypatia", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "prompt_sha256_by_order": ["a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131"], "request_sha256": "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "compiled_prompt_sha256_by_order": ["ae3c350b97daa5c6854c06d3ed1c5bfc8800493c92943b74514bed4d49857d5a", "24ffa2e91cc6fc8072e90642b47278ef16da2be9bbf05d8b4872fa44902df1e0", "0c5cc6b3a9d65f22fbf40b8e5f4b5bf10ea3acf40acf416024d37c4df7f92aff", "945fb4ad4fb6bac54e3c50412746bb58a5e332203b5e076b17b8e5612271476d", "f989215095bc58534df29645ac95b12f450bfaefa5e7431dad9bb9f9b043f7ea", "8abc5ce7d1a3f882b77d7861f043a3123a08448019880594565110d7eddd71e1"], "compiled_tokens_by_order": [309, 309, 309, 309, 309, 309], "choice": "sagan", "probs": {"darwin": 0.006249474778222577, "einstein": 0.006249474778222577, "hypatia": 0.09775823342525429, "ibn_sina": 0.004034964181507684, "sagan": 0.8185196670900704, "socrates": 0.06718818574672253}, "confidence": 0.6325775293765155, "seconds": 0.16673477066675937, "seconds_all_orders": 1.0004086240005563, "seconds_by_order": [0.1654678100021556, 0.16556828999819118, 0.16758092400050373, 0.16582875799940666, 0.17073689000244485, 0.16522595199785428], "argmax_stable": false, "spread": {"darwin": 0.004029203029902667, "einstein": 0.063652347524596, "hypatia": 0.5925813616826678, "ibn_sina": 0.0236484205979791, "sagan": 0.6565665875357383, "socrates": 0.35030922950576787}, "spread_2dp": {"darwin": 0.0, "einstein": 0.060000000000000005, "hypatia": 0.6, "ibn_sina": 0.03, "sagan": 0.66, "socrates": 0.35}, "choices_by_order": ["sagan", "sagan", "sagan", "hypatia", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.006249474778222577, "einstein": 0.006249474778222577, "hypatia": 0.09775823342525429, "ibn_sina": 0.004034964181507684, "sagan": 0.8185196670900704, "socrates": 0.06718818574672253}, {"ibn_sina": 0.012127540963192115, "hypatia": 0.07908150099536983, "sagan": 0.6621413010414658, "socrates": 0.16741553892095318, "darwin": 0.009444938398864561, "einstein": 0.06978917968015447}, {"hypatia": 0.03339253612737733, "darwin": 0.008987479173490971, "sagan": 0.522347150320018, "ibn_sina": 0.027683384779486786, "einstein": 0.048585853377662136, "socrates": 0.35900359622196476}, {"ibn_sina": 0.008929038099290482, "einstein": 0.006136832155558468, "darwin": 0.0054157353689618945, "sagan": 0.230282927714269, "hypatia": 0.6259738978100452, "socrates": 0.12326156885187499}, {"einstein": 0.013865892829096735, "sagan": 0.7570520970265224, "socrates": 0.16892115565023258, "ibn_sina": 0.004791924337959216, "darwin": 0.006972208771230668, "hypatia": 0.04839672138495844}, {"socrates": 0.008694366716196882, "ibn_sina": 0.004106928033996137, "hypatia": 0.08248985466770395, "einstein": 0.011883790302176525, "sagan": 0.8868495152500073, "darwin": 0.005975545029919188}], "prompt_tokens": 309, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.006249474778222577, "einstein": 0.006249474778222577, "hypatia": 0.09775823342525429, "ibn_sina": 0.004034964181507684, "sagan": 0.8185196670900704, "socrates": 0.06718818574672253}, "confidence": 0.6325775293765155, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.44193099980475, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "ae3c350b97daa5c6854c06d3ed1c5bfc8800493c92943b74514bed4d49857d5a", "compiled_ids_sha256": "55a7c5fbd189b1b2d5df695ec9fc001b672fd7fe50d6a7ee300faac082ac3ee1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 14.25, 17.0, 13.8125, 19.125, 16.625], "runtime_probabilities": [0.006249474361538887, 0.006249474361538887, 0.09775824099779129, 0.004034964367747307, 0.8185197114944458, 0.06718818843364716], "decide_seconds": 0.1616732599977695, "probabilities_f64_minus_runtime_max_abs": 4.440437539887654e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.012127540963192115, "hypatia": 0.07908150099536983, "sagan": 0.6621413010414658, "socrates": 0.16741553892095318, "darwin": 0.009444938398864561, "einstein": 0.06978917968015447}, "confidence": 0.4105248610438491, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.5444099997112, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "24ffa2e91cc6fc8072e90642b47278ef16da2be9bbf05d8b4872fa44902df1e0", "compiled_ids_sha256": "687ac4306df3483e9b2d71c057b95ff6abde6d6646c2e346aba92749337d1d43", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 16.75, 18.875, 17.5, 14.625, 16.625], "runtime_probabilities": [0.012127540074288845, 0.07908149808645248, 0.6621412634849548, 0.1674155294895172, 0.009444937109947205, 0.06978917121887207], "decide_seconds": 0.1618153480012552, "probabilities_f64_minus_runtime_max_abs": 3.755651101222668e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.03339253612737733, "darwin": 0.008987479173490971, "sagan": 0.522347150320018, "ibn_sina": 0.027683384779486786, "einstein": 0.048585853377662136, "socrates": 0.35900359622196476}, "confidence": 0.3809985693403313, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 167.5575539993588, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "0c5cc6b3a9d65f22fbf40b8e5f4b5bf10ea3acf40acf416024d37c4df7f92aff", "compiled_ids_sha256": "badc51827791db05c8cc51b2e7e85e2764f97366133d3f071b176052120a138b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 14.6875, 18.75, 15.8125, 16.375, 18.375], "runtime_probabilities": [0.033392533659935, 0.008987478911876678, 0.5223471522331238, 0.027683384716510773, 0.04858585074543953, 0.35900360345840454], "decide_seconds": 0.1638616929994896, "probabilities_f64_minus_runtime_max_abs": 7.236439780733406e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.008929038099290482, "einstein": 0.006136832155558468, "darwin": 0.0054157353689618945, "sagan": 0.230282927714269, "hypatia": 0.6259738978100452, "socrates": 0.12326156885187499}, "confidence": 0.44686460247367554, "choice": "hypatia"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.8027880002919, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "945fb4ad4fb6bac54e3c50412746bb58a5e332203b5e076b17b8e5612271476d", "compiled_ids_sha256": "4ee8851717638c4aadbd737007371dc1d5c01ae881489f44217879599e0ffb6a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.375, 14.25, 18.0, 19.0, 17.375], "runtime_probabilities": [0.008929037488996983, 0.0061368318274617195, 0.005415734834969044, 0.23028293251991272, 0.6259738802909851, 0.12326154857873917], "decide_seconds": 0.16208598699813592, "probabilities_f64_minus_runtime_max_abs": 2.027313582286805e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.013865892829096735, "sagan": 0.7570520970265224, "socrates": 0.16892115565023258, "ibn_sina": 0.004791924337959216, "darwin": 0.006972208771230668, "hypatia": 0.04839672138495844}, "confidence": 0.5662359512767188, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 170.71218999990379, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "f989215095bc58534df29645ac95b12f450bfaefa5e7431dad9bb9f9b043f7ea", "compiled_ids_sha256": "e5105ce9454b3cc1b9735f46cb49895d47e8684307c9d484e22722a1e07348ce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 19.0, 17.5, 13.9375, 14.3125, 16.25], "runtime_probabilities": [0.01386589277535677, 0.7570520639419556, 0.16892115771770477, 0.004791924264281988, 0.006972208619117737, 0.04839671775698662], "decide_seconds": 0.16680706099941744, "probabilities_f64_minus_runtime_max_abs": 3.3084566863905707e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.008694366716196882, "ibn_sina": 0.004106928033996137, "hypatia": 0.08248985466770395, "einstein": 0.011883790302176525, "sagan": 0.8868495152500073, "darwin": 0.005975545029919188}, "confidence": 0.7436005587095984, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.203302000009, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "8abc5ce7d1a3f882b77d7861f043a3123a08448019880594565110d7eddd71e1", "compiled_ids_sha256": "6285989cc94b20ae0b5f839c5aebba799057abbd03ed2deb13ff20954ec8fe1a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 13.875, 16.875, 14.9375, 19.25, 14.25], "runtime_probabilities": [0.008694367483258247, 0.0041069285944104195, 0.08248985558748245, 0.011883791536092758, 0.886849582195282, 0.005975545383989811], "decide_seconds": 0.16148856099971454, "probabilities_f64_minus_runtime_max_abs": 6.69452746571153e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:49Z"} {"id": "c-ibn_sina-01-dessert-01", "uid": "c054", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "prompt_sha256_by_order": ["2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d"], "request_sha256": "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "compiled_prompt_sha256_by_order": ["066735836834e07c7890804951ba5a2894b840544a1b10249ec2970aea3a1c62", "f783b341158d6a1bdbae00793d7a6096835f1bd52d229fe58519cbabe172e45f", "ede7ee9b39d29cf1e7ea56a722fe5f616f8cf1b62d53cb01af5bacb7cde9fb6c", "ff151f825d9b8ab7de0b22957e316fe6591121215d83ef608f868eb052250c79", "aca7682169f325099be485818280caadfbc9d8403c61689472e66a9ae61b1114", "3a0cf905adedfda68784b2d3929d9f06bdb65d705c7652ee9ac3948298c869eb"], "compiled_tokens_by_order": [311, 311, 311, 311, 311, 311], "choice": "sagan", "probs": {"darwin": 0.03353454754155709, "einstein": 0.08563346708303023, "hypatia": 0.0519393232833504, "ibn_sina": 0.010227459156696728, "sagan": 0.43488262945419454, "socrates": 0.383782573481171}, "confidence": 0.29987108159507303, "seconds": 0.16801963116631669, "seconds_all_orders": 1.0081177869979001, "seconds_by_order": [0.16739774599773227, 0.16583356800038018, 0.16532637199998135, 0.16520516200034763, 0.17108265799834044, 0.17327228100111824], "argmax_stable": false, "spread": {"darwin": 0.05719280138611562, "einstein": 0.40259414832829477, "hypatia": 0.18313171997810734, "ibn_sina": 0.02439192103776422, "sagan": 0.6103438722681185, "socrates": 0.4902312869104683}, "spread_2dp": {"darwin": 0.05, "einstein": 0.41, "hypatia": 0.19, "ibn_sina": 0.019999999999999997, "sagan": 0.61, "socrates": 0.49000000000000005}, "choices_by_order": ["sagan", "einstein", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.03353454754155709, "einstein": 0.08563346708303023, "hypatia": 0.0519393232833504, "ibn_sina": 0.010227459156696728, "sagan": 0.43488262945419454, "socrates": 0.383782573481171}, {"ibn_sina": 0.024178677723501327, "hypatia": 0.024178677723501327, "sagan": 0.2599452830272534, "socrates": 0.17865760609691614, "darwin": 0.027398031259586862, "einstein": 0.485641724169241}, {"hypatia": 0.05712210088183801, "darwin": 0.05041007709734682, "sagan": 0.22592229156207846, "ibn_sina": 0.018544830991981273, "einstein": 0.22592229156207846, "socrates": 0.422078407904677}, {"ibn_sina": 0.034619380194460946, "einstein": 0.08304757584094623, "darwin": 0.034619380194460946, "sagan": 0.2558045423670806, "hypatia": 0.050370900962344076, "socrates": 0.5415382204407072}, {"einstein": 0.09792649790967767, "sagan": 0.09792649790967767, "socrates": 0.49731179122888247, "ibn_sina": 0.023259581994656307, "darwin": 0.07626523325549726, "hypatia": 0.20731039770160867}, {"socrates": 0.051306933530238893, "ibn_sina": 0.01469968256860037, "hypatia": 0.045278209921549545, "einstein": 0.09585397115611256, "sagan": 0.7082703701777961, "darwin": 0.08459083264570248}], "prompt_tokens": 311, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.03353454754155709, "einstein": 0.08563346708303023, "hypatia": 0.0519393232833504, "ibn_sina": 0.010227459156696728, "sagan": 0.43488262945419454, "socrates": 0.383782573481171}, "confidence": 0.29987108159507303, "choice": "sagan"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 167.37185600140947, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "066735836834e07c7890804951ba5a2894b840544a1b10249ec2970aea3a1c62", "compiled_ids_sha256": "8b485c39edfb00d587270bcc96d96b07d12859a38e102a757f143839cbda8eb7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 16.625, 16.125, 14.5, 18.25, 18.125], "runtime_probabilities": [0.03353454917669296, 0.08563346415758133, 0.05193932726979256, 0.010227459482848644, 0.43488264083862305, 0.38378259539604187], "decide_seconds": 0.1635621050008922, "probabilities_f64_minus_runtime_max_abs": 2.1914870862449476e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.024178677723501327, "hypatia": 0.024178677723501327, "sagan": 0.2599452830272534, "socrates": 0.17865760609691614, "darwin": 0.027398031259586862, "einstein": 0.485641724169241}, "confidence": 0.28157269080021585, "choice": "einstein"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 165.810168000462, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "f783b341158d6a1bdbae00793d7a6096835f1bd52d229fe58519cbabe172e45f", "compiled_ids_sha256": "ed960119ce5f8e6565ff801303fdbcec4f6bdb265f371600327c44bafaa8469c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 15.25, 17.625, 17.25, 15.375, 18.25], "runtime_probabilities": [0.024178676307201385, 0.024178676307201385, 0.2599453032016754, 0.1786576211452484, 0.027398033067584038, 0.4856417179107666], "decide_seconds": 0.1620239270014281, "probabilities_f64_minus_runtime_max_abs": 2.0174422021845828e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.05712210088183801, "darwin": 0.05041007709734682, "sagan": 0.22592229156207846, "ibn_sina": 0.018544830991981273, "einstein": 0.22592229156207846, "socrates": 0.422078407904677}, "confidence": 0.20509126205725492, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 165.30266199697508, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "ede7ee9b39d29cf1e7ea56a722fe5f616f8cf1b62d53cb01af5bacb7cde9fb6c", "compiled_ids_sha256": "3e0522a64fe8a366e627d965d820c44e3c949ca7db48f92f8d91ac37b00bbdf3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 16.125, 17.625, 15.125, 17.625, 18.25], "runtime_probabilities": [0.05712209641933441, 0.05041007697582245, 0.22592228651046753, 0.01854483038187027, 0.22592228651046753, 0.422078400850296], "decide_seconds": 0.16158520099997986, "probabilities_f64_minus_runtime_max_abs": 7.054380968263274e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.034619380194460946, "einstein": 0.08304757584094623, "darwin": 0.034619380194460946, "sagan": 0.2558045423670806, "hypatia": 0.050370900962344076, "socrates": 0.5415382204407072}, "confidence": 0.2906711270903142, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 165.18240200093715, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "ff151f825d9b8ab7de0b22957e316fe6591121215d83ef608f868eb052250c79", "compiled_ids_sha256": "c5abb226e29b161b96dee1f321148a8bfae64ee1d02ae5e8dc7ab6a26bf07807", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 16.5, 15.625, 17.625, 16.0, 18.375], "runtime_probabilities": [0.03461937978863716, 0.08304757624864578, 0.03461937978863716, 0.25580456852912903, 0.050370898097753525, 0.5415382385253906], "decide_seconds": 0.16148486100064474, "probabilities_f64_minus_runtime_max_abs": 2.6162048416544792e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.09792649790967767, "sagan": 0.09792649790967767, "socrates": 0.49731179122888247, "ibn_sina": 0.023259581994656307, "darwin": 0.07626523325549726, "hypatia": 0.20731039770160867}, "confidence": 0.21171026261428538, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 171.05477799850632, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "aca7682169f325099be485818280caadfbc9d8403c61689472e66a9ae61b1114", "compiled_ids_sha256": "e83fedc7e71580c6254aa4d6e7eeadd714de9f75b82ed69cb74b9b88884c3021", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 16.5, 18.125, 15.0625, 16.25, 17.25], "runtime_probabilities": [0.09792648255825043, 0.09792648255825043, 0.4973117709159851, 0.023259581997990608, 0.07626523077487946, 0.20731037855148315], "decide_seconds": 0.16729654600203503, "probabilities_f64_minus_runtime_max_abs": 2.0312897364149762e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.051306933530238893, "ibn_sina": 0.01469968256860037, "hypatia": 0.045278209921549545, "einstein": 0.09585397115611256, "sagan": 0.7082703701777961, "darwin": 0.08459083264570248}, "confidence": 0.42372293459287513, "choice": "sagan"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 173.2485610009462, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "3a0cf905adedfda68784b2d3929d9f06bdb65d705c7652ee9ac3948298c869eb", "compiled_ids_sha256": "84a654325c4190581d7415981d6d5c64d1878d9d2ff32034c6ae1e2cbe18310d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 14.75, 15.875, 16.625, 18.625, 16.5], "runtime_probabilities": [0.05130693316459656, 0.014699681662023067, 0.04527820646762848, 0.09585396945476532, 0.7082703113555908, 0.08459082990884781], "decide_seconds": 0.16939712099701865, "probabilities_f64_minus_runtime_max_abs": 5.882220532882343e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:50Z"} {"id": "c-ibn_sina-01-dessert-02", "uid": "c055", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "prompt_sha256_by_order": ["a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e"], "request_sha256": "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "compiled_prompt_sha256_by_order": ["aa541bf54e4da10cb7233c63ec2579c3ba492ff5efbefd219bec4731f7f5d42b", "719ff064b126def9a53c8d0e1237002376a4bec9812d53dafb458ae3d6e15ff6", "38403e234882e32610254e467c79c997b2faac2df87dc46093553a8908056822", "61119adef9837a7be193e2b75f6706c60b7e71f5a804a76e053a6ca7453326c2", "2b1dbb917cf71dfd3a50147a103cc32e984ad22b21f79511951e9458a5e6fa2b", "cb95ff037757ccada52a6794bc80b016c20ce88676cada73060577fd944f2636"], "compiled_tokens_by_order": [323, 323, 323, 323, 323, 323], "choice": "ibn_sina", "probs": {"darwin": 0.004841551099422038, "einstein": 0.0025914955577243277, "hypatia": 0.0033275461633514108, "ibn_sina": 0.9821412572286408, "sagan": 0.0037706037875101044, "socrates": 0.0033275461633514108}, "confidence": 0.9341693442679537, "seconds": 0.1737646403338052, "seconds_all_orders": 1.0425878420028312, "seconds_by_order": [0.17081948000122793, 0.1705236320012773, 0.17406534499968984, 0.1784013610013062, 0.17814846300097997, 0.17062956099834992], "argmax_stable": true, "spread": {"darwin": 0.027045070036879018, "einstein": 0.06284992562950249, "hypatia": 0.09221065906805878, "ibn_sina": 0.509538484811257, "sagan": 0.01521044909413074, "socrates": 0.3243843685000363}, "spread_2dp": {"darwin": 0.03, "einstein": 0.06, "hypatia": 0.09, "ibn_sina": 0.51, "sagan": 0.02, "socrates": 0.33}, "choices_by_order": ["ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.004841551099422038, "einstein": 0.0025914955577243277, "hypatia": 0.0033275461633514108, "ibn_sina": 0.9821412572286408, "sagan": 0.0037706037875101044, "socrates": 0.0033275461633514108}, {"ibn_sina": 0.7892469960242797, "hypatia": 0.050454872419651034, "sagan": 0.00773750528708068, "socrates": 0.06478533858256925, "darwin": 0.030602427054407695, "einstein": 0.05717286063201164}, {"hypatia": 0.007884062618286183, "darwin": 0.0065361174801168535, "sagan": 0.0025595803818343407, "ibn_sina": 0.9700458434897147, "einstein": 0.007884062618286183, "socrates": 0.0050903334117617036}, {"ibn_sina": 0.47681802230332176, "einstein": 0.06453030210074157, "darwin": 0.019680630200445248, "sagan": 0.017368095192805775, "hypatia": 0.09389103553929785, "socrates": 0.3277119146633877}, {"einstein": 0.006554132874498994, "sagan": 0.002411130739234069, "socrates": 0.005784001960873233, "ibn_sina": 0.972719565082447, "darwin": 0.007426805527932966, "hypatia": 0.005104363815013855}, {"socrates": 0.004567736826739321, "ibn_sina": 0.9863565071145788, "hypatia": 0.0016803764712390787, "einstein": 0.0016803764712390787, "sagan": 0.0021576460986750348, "darwin": 0.0035573570175286774}], "prompt_tokens": 323, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.004841551099422038, "einstein": 0.0025914955577243277, "hypatia": 0.0033275461633514108, "ibn_sina": 0.9821412572286408, "sagan": 0.0037706037875101044, "socrates": 0.0033275461633514108}, "confidence": 0.9341693442679537, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.7926699964446, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "aa541bf54e4da10cb7233c63ec2579c3ba492ff5efbefd219bec4731f7f5d42b", "compiled_ids_sha256": "6d3e88b7607381a1857be3378d363a3a49ad70080bd24153eee4a6f79fe0c29c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.6875, 13.9375, 19.625, 14.0625, 13.9375], "runtime_probabilities": [0.004841551184654236, 0.0025914956349879503, 0.003327546175569296, 0.9821412563323975, 0.003770604031160474, 0.003327546175569296], "decide_seconds": 0.16685864000100992, "probabilities_f64_minus_runtime_max_abs": 8.962433017956073e-10}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.7892469960242797, "hypatia": 0.050454872419651034, "sagan": 0.00773750528708068, "socrates": 0.06478533858256925, "darwin": 0.030602427054407695, "einstein": 0.05717286063201164}, "confidence": 0.5408352368473276, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.5004219984403, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "719ff064b126def9a53c8d0e1237002376a4bec9812d53dafb458ae3d6e15ff6", "compiled_ids_sha256": "2e5335910a403d03f98920ce9cac22c9f0273eeda65f8ae3961984e75c23f585", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 16.625, 14.75, 16.875, 16.125, 16.75], "runtime_probabilities": [0.7892469763755798, 0.050454869866371155, 0.007737505715340376, 0.06478533893823624, 0.030602427199482918, 0.05717286467552185], "decide_seconds": 0.16669296100008069, "probabilities_f64_minus_runtime_max_abs": 1.9648699889351917e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.007884062618286183, "darwin": 0.0065361174801168535, "sagan": 0.0025595803818343407, "ibn_sina": 0.9700458434897147, "einstein": 0.007884062618286183, "socrates": 0.0050903334117617036}, "confidence": 0.8990386826975705, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 174.04247499871417, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "38403e234882e32610254e467c79c997b2faac2df87dc46093553a8908056822", "compiled_ids_sha256": "b3ae8862f8055606217f15284cb54126369ae13606edd1ab769fae010e5b8581", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.875, 13.9375, 19.875, 15.0625, 14.625], "runtime_probabilities": [0.007884063757956028, 0.006536118220537901, 0.00255958060733974, 0.9700459241867065, 0.007884063757956028, 0.005090333987027407], "decide_seconds": 0.17029584299962153, "probabilities_f64_minus_runtime_max_abs": 8.069699186652457e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.47681802230332176, "einstein": 0.06453030210074157, "darwin": 0.019680630200445248, "sagan": 0.017368095192805775, "hypatia": 0.09389103553929785, "socrates": 0.3277119146633877}, "confidence": 0.29376140254067395, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 178.3780609985115, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "61119adef9837a7be193e2b75f6706c60b7e71f5a804a76e053a6ca7453326c2", "compiled_ids_sha256": "f81a84892bebe7604274c8a613f81476a0366788d4ddfa128180c78947b642c4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.625, 16.625, 15.4375, 15.3125, 17.0, 18.25], "runtime_probabilities": [0.4768180251121521, 0.06453029811382294, 0.019680630415678024, 0.017368094995617867, 0.09389103204011917, 0.32771191000938416], "decide_seconds": 0.1745759210025426, "probabilities_f64_minus_runtime_max_abs": 4.654003571413767e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.006554132874498994, "sagan": 0.002411130739234069, "socrates": 0.005784001960873233, "ibn_sina": 0.972719565082447, "darwin": 0.007426805527932966, "hypatia": 0.005104363815013855}, "confidence": 0.9064921917136713, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 178.12498300190782, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "2b1dbb917cf71dfd3a50147a103cc32e984ad22b21f79511951e9458a5e6fa2b", "compiled_ids_sha256": "486109abcac25cd8a2cd9f0d4081419470451170e163118482ebcf0a6ddca8fd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 13.75, 14.625, 19.75, 14.875, 14.5], "runtime_probabilities": [0.006554132793098688, 0.0024111308157444, 0.005784001667052507, 0.9727195501327515, 0.00742680486291647, 0.005104363895952702], "decide_seconds": 0.1742793730008998, "probabilities_f64_minus_runtime_max_abs": 1.4949695481547565e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.004567736826739321, "ibn_sina": 0.9863565071145788, "hypatia": 0.0016803764712390787, "einstein": 0.0016803764712390787, "sagan": 0.0021576460986750348, "darwin": 0.0035573570175286774}, "confidence": 0.9481294653248441, "choice": "ibn_sina"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.60666100223898, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "cb95ff037757ccada52a6794bc80b016c20ce88676cada73060577fd944f2636", "compiled_ids_sha256": "ee9ff1d8debac08d3d0dc0979ef0b33d378160d30a736188f3070566f29f0134", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 19.75, 13.375, 13.375, 13.625, 14.125], "runtime_probabilities": [0.004567737225443125, 0.9863564968109131, 0.0016803765902295709, 0.0016803765902295709, 0.002157646231353283, 0.003557357005774975], "decide_seconds": 0.16679666099662427, "probabilities_f64_minus_runtime_max_abs": 1.0303665698430109e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:51Z"} {"id": "c-ibn_sina-01-dessert-02", "uid": "c056", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "prompt_sha256_by_order": ["ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595"], "request_sha256": "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "compiled_prompt_sha256_by_order": ["725690d1f22c93d0340b285452fdb898cbeffbf60adbcee2a6cf9213e82abbc0", "b8aac5aa144a3a3b9b2042865a2d2015541abec27b5967e49c0db4498c4881a8", "ef9e5a25d1357e0028a89e6361af2e9194a3bac6232a11e3d3d52552c9527f95", "e919be9065d309bf82ab619c6d58503f847ac29334d5c3dfc4aee4836de71670", "2a5d84bd4af93d07480a033a6331117b9c81bae0d818e013fc68eebed536bfd5", "ebf830a50a0084bf727dee0ecca59607345f1d5b5ff40a355f83f10100363532"], "compiled_tokens_by_order": [342, 342, 342, 342, 342, 342], "choice": "socrates", "probs": {"darwin": 0.009979564645855131, "einstein": 0.004160101473783409, "hypatia": 0.1883005707473848, "ibn_sina": 0.028876827783846305, "sagan": 0.023939731073489384, "socrates": 0.744743204275641}, "confidence": 0.556643050876081, "seconds": 0.17403176183385463, "seconds_all_orders": 1.0441905710031278, "seconds_by_order": [0.17537460400126292, 0.17446634199950495, 0.17718323100052658, 0.1717410530000052, 0.17247640699861222, 0.17294893400321598], "argmax_stable": false, "spread": {"darwin": 0.02022440523449329, "einstein": 0.004235567967558255, "hypatia": 0.6731721044914534, "ibn_sina": 0.034936451962036146, "sagan": 0.09048071269736228, "socrates": 0.6281555557715903}, "spread_2dp": {"darwin": 0.02, "einstein": 0.01, "hypatia": 0.6699999999999999, "ibn_sina": 0.03, "sagan": 0.09, "socrates": 0.62}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.009979564645855131, "einstein": 0.004160101473783409, "hypatia": 0.1883005707473848, "ibn_sina": 0.028876827783846305, "sagan": 0.023939731073489384, "socrates": 0.744743204275641}, {"ibn_sina": 0.04338203289286428, "hypatia": 0.32055227473102893, "sagan": 0.021813856055106214, "socrates": 0.5988704914119479, "darwin": 0.012429160521352166, "einstein": 0.002952184387700498}, {"hypatia": 0.25337700446561673, "darwin": 0.004640761716501424, "sagan": 0.0071877523552587525, "ibn_sina": 0.038856622119091945, "einstein": 0.0071877523552587525, "socrates": 0.6887501069882723}, {"ibn_sina": 0.03032235454461745, "einstein": 0.004103684440696793, "darwin": 0.009247795642855801, "sagan": 0.09339942736719692, "hypatia": 0.2538859662007318, "socrates": 0.6090407718039011}, {"einstein": 0.005122503773631628, "sagan": 0.0029187146698346415, "socrates": 0.11658764850405064, "ibn_sina": 0.008445580930828138, "darwin": 0.005452876882816693, "hypatia": 0.8614726752388382}, {"socrates": 0.26732573776413415, "ibn_sina": 0.014167745062743759, "hypatia": 0.6412811077639087, "einstein": 0.00590599482114558, "sagan": 0.04645424763707317, "darwin": 0.024865166950994715}], "prompt_tokens": 342, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.009979564645855131, "einstein": 0.004160101473783409, "hypatia": 0.1883005707473848, "ibn_sina": 0.028876827783846305, "sagan": 0.023939731073489384, "socrates": 0.744743204275641}, "confidence": 0.556643050876081, "choice": "socrates"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 175.34763499861583, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "725690d1f22c93d0340b285452fdb898cbeffbf60adbcee2a6cf9213e82abbc0", "compiled_ids_sha256": "94b95ec2aa38d8983aa6f7c9626041fbc139ec74d98175507b3c7f676e48c4ab", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.1875, 18.0, 16.125, 15.9375, 19.375], "runtime_probabilities": [0.00997956469655037, 0.004160101525485516, 0.18830056488513947, 0.028876828029751778, 0.02393973246216774, 0.7447432279586792], "decide_seconds": 0.1712232869977015, "probabilities_f64_minus_runtime_max_abs": 2.368303819721973e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04338203289286428, "hypatia": 0.32055227473102893, "sagan": 0.021813856055106214, "socrates": 0.5988704914119479, "darwin": 0.012429160521352166, "einstein": 0.002952184387700498}, "confidence": 0.46251837281238395, "choice": "socrates"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 174.44102199806366, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "b8aac5aa144a3a3b9b2042865a2d2015541abec27b5967e49c0db4498c4881a8", "compiled_ids_sha256": "3f3a2569cd4a2ce0f1e66875c36c96917d14aaaa302d580d85aae029b59bcd14", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 18.625, 15.9375, 19.25, 15.375, 13.9375], "runtime_probabilities": [0.043382033705711365, 0.3205522894859314, 0.021813856437802315, 0.5988704562187195, 0.012429160065948963, 0.002952184295281768], "decide_seconds": 0.17034900300131994, "probabilities_f64_minus_runtime_max_abs": 3.51932284425871e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.25337700446561673, "darwin": 0.004640761716501424, "sagan": 0.0071877523552587525, "ibn_sina": 0.038856622119091945, "einstein": 0.0071877523552587525, "socrates": 0.6887501069882723}, "confidence": 0.5385772369231052, "choice": "socrates"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 177.15887099984684, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "ef9e5a25d1357e0028a89e6361af2e9194a3bac6232a11e3d3d52552c9527f95", "compiled_ids_sha256": "864494f70d217d88f3658a0a30829020dc43e406f70060ad983ecbf868de4275", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 14.5, 14.9375, 16.625, 14.9375, 19.5], "runtime_probabilities": [0.2533770203590393, 0.00464076129719615, 0.007187752053141594, 0.0388566218316555, 0.007187752053141594, 0.6887500882148743], "decide_seconds": 0.1731439320028585, "probabilities_f64_minus_runtime_max_abs": 1.8773398058336e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03032235454461745, "einstein": 0.004103684440696793, "darwin": 0.009247795642855801, "sagan": 0.09339942736719692, "hypatia": 0.2538859662007318, "socrates": 0.6090407718039011}, "confidence": 0.417692575656503, "choice": "socrates"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 171.71792300359812, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "e919be9065d309bf82ab619c6d58503f847ac29334d5c3dfc4aee4836de71670", "compiled_ids_sha256": "a58249ff3fdeebc3c38396447d68f6b88df0bc950b26791659c8521322bd44f1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 14.25, 15.0625, 17.375, 18.375, 19.25], "runtime_probabilities": [0.030322352424263954, 0.0041036843322217464, 0.0092477947473526, 0.09339942783117294, 0.253885954618454, 0.6090407371520996], "decide_seconds": 0.1677375530016434, "probabilities_f64_minus_runtime_max_abs": 3.465180153927605e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.005122503773631628, "sagan": 0.0029187146698346415, "socrates": 0.11658764850405064, "ibn_sina": 0.008445580930828138, "darwin": 0.005452876882816693, "hypatia": 0.8614726752388382}, "confidence": 0.7255176996817312, "choice": "hypatia"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 172.45102699962445, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "2a5d84bd4af93d07480a033a6331117b9c81bae0d818e013fc68eebed536bfd5", "compiled_ids_sha256": "7855a162de97f7e775b36190545344eebc7e1d0de7bb34c8275ef5493cc8b43c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 13.9375, 17.625, 15.0, 14.5625, 19.625], "runtime_probabilities": [0.005122503265738487, 0.0029187146574258804, 0.11658764630556107, 0.008445581421256065, 0.005452876444905996, 0.8614726662635803], "decide_seconds": 0.16837482799746795, "probabilities_f64_minus_runtime_max_abs": 8.975257870247333e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.26732573776413415, "ibn_sina": 0.014167745062743759, "hypatia": 0.6412811077639087, "einstein": 0.00590599482114558, "sagan": 0.04645424763707317, "darwin": 0.024865166950994715}, "confidence": 0.4627345461089033, "choice": "hypatia"}}, "usage": {"input_tokens": 342, "output_tokens": 1}, "latency_ms": 172.9225140006747, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 342, "compiled_prompt_sha256": "ebf830a50a0084bf727dee0ecca59607345f1d5b5ff40a355f83f10100363532", "compiled_ids_sha256": "735e6ae15336ca10f18bed8cf9f335970bb12077fe9c321d6e337aa2879ee706", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.375, 15.4375, 19.25, 14.5625, 16.625, 16.0], "runtime_probabilities": [0.2673257291316986, 0.014167743735015392, 0.6412810683250427, 0.005905994214117527, 0.04645424708724022, 0.02486516535282135], "decide_seconds": 0.16827216899764608, "probabilities_f64_minus_runtime_max_abs": 3.9438865950813806e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:53Z"} {"id": "c-ibn_sina-01-dessert-02", "uid": "c057", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "prompt_sha256_by_order": ["fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e"], "request_sha256": "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "compiled_prompt_sha256_by_order": ["319fd600c27e869175120d52429454dc20b5548f94e0e370905b96ceb450ebfc", "fc07f5d1d7837dc5f780d230c04d7ba189bd12134f043747eb2c9bb060d658e3", "b1de02c44ffe4e7cb5658167df659bd0c9653ca47e976d09991fbeec1f0d4408", "cb3a577fa0c69484f0047139df1cc7529c904331ef0dd79451cd3e3aa328e91d", "6dba88eb5598bf7f49e6233f1684908940026c2b713555bb23dc984a684cec8b", "233a7140192e50df93982d534c857e749e0fe6edc924badc6e732211857a206a"], "compiled_tokens_by_order": [339, 339, 339, 339, 339, 339], "choice": "darwin", "probs": {"darwin": 0.6988862658277547, "einstein": 0.03942846507021263, "hypatia": 0.01862468812508585, "ibn_sina": 0.09458397073597778, "sagan": 0.0834700612086524, "socrates": 0.06500654903231659}, "confidence": 0.40836312850761813, "seconds": 0.17084519933268894, "seconds_all_orders": 1.0250711959961336, "seconds_by_order": [0.17131622599845286, 0.17076350999923307, 0.1705795109992323, 0.17058316099792137, 0.17060582100020838, 0.17122296700108564], "argmax_stable": false, "spread": {"darwin": 0.6203873694754607, "einstein": 0.8045142047057412, "hypatia": 0.05848527661073477, "ibn_sina": 0.24122308489582606, "sagan": 0.07812814427185803, "socrates": 0.2244740179306951}, "spread_2dp": {"darwin": 0.62, "einstein": 0.7999999999999999, "hypatia": 0.06, "ibn_sina": 0.24000000000000002, "sagan": 0.07, "socrates": 0.22}, "choices_by_order": ["darwin", "einstein", "darwin", "ibn_sina", "einstein", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.6988862658277547, "einstein": 0.03942846507021263, "hypatia": 0.01862468812508585, "ibn_sina": 0.09458397073597778, "sagan": 0.0834700612086524, "socrates": 0.06500654903231659}, {"ibn_sina": 0.2204261411440733, "hypatia": 0.010974371359204404, "sagan": 0.008029021604258342, "socrates": 0.043404480711475596, "darwin": 0.11798561119170525, "einstein": 0.5991803739892831}, {"hypatia": 0.022822460592679623, "darwin": 0.4879657235544313, "sagan": 0.022822460592679623, "ibn_sina": 0.139804520501698, "einstein": 0.23049868669118168, "socrates": 0.09608614806732972}, {"ibn_sina": 0.2832405162945191, "einstein": 0.13379334627862716, "darwin": 0.22058793588772047, "sagan": 0.04921982146143864, "hypatia": 0.06319950176131998, "socrates": 0.2499588783163747}, {"einstein": 0.8439426697759538, "sagan": 0.005341916936794364, "socrates": 0.0254848603856796, "ibn_sina": 0.04201743139869303, "darwin": 0.078498896352294, "hypatia": 0.004714225150585216}, {"socrates": 0.03652311604197307, "ibn_sina": 0.05314082027617512, "hypatia": 0.009234489897582323, "einstein": 0.18547968780469323, "sagan": 0.06823416389824413, "darwin": 0.6473877220813321}], "prompt_tokens": 339, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.6988862658277547, "einstein": 0.03942846507021263, "hypatia": 0.01862468812508585, "ibn_sina": 0.09458397073597778, "sagan": 0.0834700612086524, "socrates": 0.06500654903231659}, "confidence": 0.40836312850761813, "choice": "darwin"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 171.28896600115695, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "319fd600c27e869175120d52429454dc20b5548f94e0e370905b96ceb450ebfc", "compiled_ids_sha256": "7cd4b48291d7d001e70cad1045450e5ebc69a94b89e94cfc8233e6b63d6d448e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 16.5, 15.75, 17.375, 17.25, 17.0], "runtime_probabilities": [0.6988863348960876, 0.0394284687936306, 0.018624689429998398, 0.09458397328853607, 0.08347006887197495, 0.0650065541267395], "decide_seconds": 0.16690050900069764, "probabilities_f64_minus_runtime_max_abs": 6.906833294184622e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.2204261411440733, "hypatia": 0.010974371359204404, "sagan": 0.008029021604258342, "socrates": 0.043404480711475596, "darwin": 0.11798561119170525, "einstein": 0.5991803739892831}, "confidence": 0.3766993938145382, "choice": "einstein"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.73835100018186, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "fc07f5d1d7837dc5f780d230c04d7ba189bd12134f043747eb2c9bb060d658e3", "compiled_ids_sha256": "81f8dcad8d99c88fc716915bfa3dcb041f1df8ebdd94e5ff0b44d693f0fe059a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.0, 15.0, 14.6875, 16.375, 17.375, 19.0], "runtime_probabilities": [0.22042615711688995, 0.010974370874464512, 0.008029021322727203, 0.0434044785797596, 0.11798559874296188, 0.5991803407669067], "decide_seconds": 0.16672313100207248, "probabilities_f64_minus_runtime_max_abs": 3.322237640546177e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.022822460592679623, "darwin": 0.4879657235544313, "sagan": 0.022822460592679623, "ibn_sina": 0.139804520501698, "einstein": 0.23049868669118168, "socrates": 0.09608614806732972}, "confidence": 0.24037385320180404, "choice": "darwin"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.5556320011965, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "b1de02c44ffe4e7cb5658167df659bd0c9653ca47e976d09991fbeec1f0d4408", "compiled_ids_sha256": "c2db1c0fc322cab49cd6dd4fde364dfba39de91c1e5ae1c48a6063304bbfc372", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 19.0, 15.9375, 17.75, 18.25, 17.375], "runtime_probabilities": [0.022822458297014236, 0.4879656732082367, 0.022822458297014236, 0.13980451226234436, 0.23049867153167725, 0.09608612954616547], "decide_seconds": 0.16659557299863081, "probabilities_f64_minus_runtime_max_abs": 5.034619460220213e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.2832405162945191, "einstein": 0.13379334627862716, "darwin": 0.22058793588772047, "sagan": 0.04921982146143864, "hypatia": 0.06319950176131998, "socrates": 0.2499588783163747}, "confidence": 0.09076542748472682, "choice": "ibn_sina"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.55991099914536, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "cb3a577fa0c69484f0047139df1cc7529c904331ef0dd79451cd3e3aa328e91d", "compiled_ids_sha256": "51611cd39337ce5f4a25d3628163e7c9e812a7adbe100e8a34f3da1348be6203", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.25, 17.5, 18.0, 16.5, 16.75, 18.125], "runtime_probabilities": [0.28324052691459656, 0.13379335403442383, 0.22058795392513275, 0.04921982064843178, 0.06319950520992279, 0.24995888769626617], "decide_seconds": 0.16658781200021622, "probabilities_f64_minus_runtime_max_abs": 1.8037412286009413e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.8439426697759538, "sagan": 0.005341916936794364, "socrates": 0.0254848603856796, "ibn_sina": 0.04201743139869303, "darwin": 0.078498896352294, "hypatia": 0.004714225150585216}, "confidence": 0.6523789012756078, "choice": "einstein"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 170.58232100316673, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "6dba88eb5598bf7f49e6233f1684908940026c2b713555bb23dc984a684cec8b", "compiled_ids_sha256": "47f5ea364455f61ec4c0c6af4a5f10bfe2973af14c53619a907bdc142c2a2e51", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 14.5625, 16.125, 16.625, 17.25, 14.4375], "runtime_probabilities": [0.8439427018165588, 0.005341916810721159, 0.025484861806035042, 0.04201743006706238, 0.07849889248609543, 0.004714225418865681], "decide_seconds": 0.16658221199759282, "probabilities_f64_minus_runtime_max_abs": 3.204060505801465e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.03652311604197307, "ibn_sina": 0.05314082027617512, "hypatia": 0.009234489897582323, "einstein": 0.18547968780469323, "sagan": 0.06823416389824413, "darwin": 0.6473877220813321}, "confidence": 0.3875913640289943, "choice": "darwin"}}, "usage": {"input_tokens": 339, "output_tokens": 1}, "latency_ms": 171.19836699930602, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 339, "compiled_prompt_sha256": "233a7140192e50df93982d534c857e749e0fe6edc924badc6e732211857a206a", "compiled_ids_sha256": "1ea83d5358ad7332ffb1aa8de7a81adc06c9282cc30823f1ace50ed669c3eb96", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 16.5, 14.75, 17.75, 16.75, 19.0], "runtime_probabilities": [0.03652311488986015, 0.05314081534743309, 0.009234488941729069, 0.18547968566417694, 0.0682341530919075, 0.647387683391571], "decide_seconds": 0.16712552800163394, "probabilities_f64_minus_runtime_max_abs": 3.868976106158328e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:54Z"} {"id": "c-ibn_sina-01-dessert-03", "uid": "c058", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "prompt_sha256_by_order": ["a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c"], "request_sha256": "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "compiled_prompt_sha256_by_order": ["31e37b8f5f725ef736f2d61439d483950b0c0100444ccf92ee30ba189e9af0ef", "7064ccd748648c0f2c9753d94b5d60388cc828d9f69d0e10739045a782a9026c", "b18c386c42d75a2b1507d314bf26f7105eede4ceeefde8f9b570ab7db106cdd1", "aa2c4e11f21d36ddf82ad40b63631513ad43985be9fcfa74dc72ea6a452c492b", "6012266a2fa2c2dd0247aced474f41b0d142b78c9f2081c7fb744f049f787e5d", "a7a04181e9aebfc106fb6edbfdd1814d937a130ef09de73af59f3cc5ec826651"], "compiled_tokens_by_order": [351, 351, 351, 351, 351, 351], "choice": "hypatia", "probs": {"darwin": 0.008491729750069864, "einstein": 0.0040112091088488205, "hypatia": 0.5953162156652845, "ibn_sina": 0.07110038269421554, "sagan": 0.1505195113448218, "socrates": 0.17056095143675945}, "confidence": 0.36037045876191387, "seconds": 0.17170335800013467, "seconds_all_orders": 1.030220148000808, "seconds_by_order": [0.17138902600345318, 0.17087328900015564, 0.17128065599899855, 0.17122517700045137, 0.1740875449977466, 0.17136445500000264], "argmax_stable": false, "spread": {"darwin": 0.0637440775506163, "einstein": 0.07895600699430211, "hypatia": 0.5151990591913415, "ibn_sina": 0.12820625529176594, "sagan": 0.6323882186279954, "socrates": 0.18220099378575885}, "spread_2dp": {"darwin": 0.060000000000000005, "einstein": 0.08, "hypatia": 0.52, "ibn_sina": 0.13, "sagan": 0.64, "socrates": 0.18000000000000002}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.008491729750069864, "einstein": 0.0040112091088488205, "hypatia": 0.5953162156652845, "ibn_sina": 0.07110038269421554, "sagan": 0.1505195113448218, "socrates": 0.17056095143675945}, {"ibn_sina": 0.08150912955129543, "hypatia": 0.4139375152172762, "sagan": 0.22156478571280053, "socrates": 0.17255482861417704, "darwin": 0.07193155436138472, "einstein": 0.03850218654306619}, {"hypatia": 0.4199374487694894, "darwin": 0.008187476810768429, "sagan": 0.15448635398028096, "ibn_sina": 0.13633372887918532, "einstein": 0.08269058651817558, "socrates": 0.19836440504210018}, {"ibn_sina": 0.054879245490452415, "einstein": 0.003734579523873476, "darwin": 0.010806364163467195, "sagan": 0.19154738800842008, "hypatia": 0.6685660767553998, "socrates": 0.07046634605838702}, {"einstein": 0.011699243012654107, "sagan": 0.013256979121841877, "socrates": 0.059413658436058656, "ibn_sina": 0.03603620544716373, "darwin": 0.059413658436058656, "hypatia": 0.8201802555462229}, {"socrates": 0.01616341125634133, "ibn_sina": 0.00812747358741938, "hypatia": 0.3049811963548814, "einstein": 0.0055859254602901405, "sagan": 0.6456451977498373, "darwin": 0.01949679559123049}], "prompt_tokens": 351, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.008491729750069864, "einstein": 0.0040112091088488205, "hypatia": 0.5953162156652845, "ibn_sina": 0.07110038269421554, "sagan": 0.1505195113448218, "socrates": 0.17056095143675945}, "confidence": 0.36037045876191387, "choice": "hypatia"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 171.36099599883892, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "31e37b8f5f725ef736f2d61439d483950b0c0100444ccf92ee30ba189e9af0ef", "compiled_ids_sha256": "a7851da9192bd84a58d364630f143cf2398c247083217825c24b2c65fd8f0f9e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 14.375, 19.375, 17.25, 18.0, 18.125], "runtime_probabilities": [0.008491729386150837, 0.00401120912283659, 0.5953162312507629, 0.0711003839969635, 0.1505195051431656, 0.17056095600128174], "decide_seconds": 0.16691759999957867, "probabilities_f64_minus_runtime_max_abs": 1.5585478463364666e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.08150912955129543, "hypatia": 0.4139375152172762, "sagan": 0.22156478571280053, "socrates": 0.17255482861417704, "darwin": 0.07193155436138472, "einstein": 0.03850218654306619}, "confidence": 0.1509580460422718, "choice": "hypatia"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 170.84913899816456, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "7064ccd748648c0f2c9753d94b5d60388cc828d9f69d0e10739045a782a9026c", "compiled_ids_sha256": "2324df4a215a5745c4707c4d8f8edf1135993030acac8d65f6e7a78ccd5b039b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 18.5, 17.875, 17.625, 16.75, 16.125], "runtime_probabilities": [0.08150912821292877, 0.4139375388622284, 0.22156479954719543, 0.1725548356771469, 0.07193154841661453, 0.03850218653678894], "decide_seconds": 0.16671378099999856, "probabilities_f64_minus_runtime_max_abs": 2.364495221840457e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.4199374487694894, "darwin": 0.008187476810768429, "sagan": 0.15448635398028096, "ibn_sina": 0.13633372887918532, "einstein": 0.08269058651817558, "socrates": 0.19836440504210018}, "confidence": 0.1679156540330523, "choice": "hypatia"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 171.2562960019568, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "b18c386c42d75a2b1507d314bf26f7105eede4ceeefde8f9b570ab7db106cdd1", "compiled_ids_sha256": "abf1efbfa2153e30dab711df02d4d549b99c25d5d3d9bb83ebf6cff664d7c286", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.0, 15.0625, 18.0, 17.875, 17.375, 18.25], "runtime_probabilities": [0.4199374318122864, 0.008187476545572281, 0.15448635816574097, 0.13633371889591217, 0.08269058167934418, 0.19836440682411194], "decide_seconds": 0.16715230800036807, "probabilities_f64_minus_runtime_max_abs": 1.6957203030631973e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.054879245490452415, "einstein": 0.003734579523873476, "darwin": 0.010806364163467195, "sagan": 0.19154738800842008, "hypatia": 0.6685660767553998, "socrates": 0.07046634605838702}, "confidence": 0.4409120911386396, "choice": "hypatia"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 171.1997069978679, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "aa2c4e11f21d36ddf82ad40b63631513ad43985be9fcfa74dc72ea6a452c492b", "compiled_ids_sha256": "12e049f0c86c059791799a19d3065dccdb225bd542b964eb7bfd3b24a9e6646e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 14.3125, 15.375, 18.25, 19.5, 17.25], "runtime_probabilities": [0.05487924441695213, 0.0037345795426517725, 0.010806364007294178, 0.19154739379882812, 0.6685660481452942, 0.07046633958816528], "decide_seconds": 0.16694387899769936, "probabilities_f64_minus_runtime_max_abs": 2.8610105617943304e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.011699243012654107, "sagan": 0.013256979121841877, "socrates": 0.059413658436058656, "ibn_sina": 0.03603620544716373, "darwin": 0.059413658436058656, "hypatia": 0.8201802555462229}, "confidence": 0.5941571276578428, "choice": "hypatia"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 174.0627350009163, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "6012266a2fa2c2dd0247aced474f41b0d142b78c9f2081c7fb744f049f787e5d", "compiled_ids_sha256": "6a623731292bdcae5d1ab665f7d81dcf414d34e8b7dcda00141decf2cd544432", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 15.375, 16.875, 16.375, 16.875, 19.5], "runtime_probabilities": [0.01169924158602953, 0.013256978243589401, 0.05941365659236908, 0.03603620454668999, 0.05941365659236908, 0.8201802372932434], "decide_seconds": 0.16992580700025428, "probabilities_f64_minus_runtime_max_abs": 1.8252979461230723e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.01616341125634133, "ibn_sina": 0.00812747358741938, "hypatia": 0.3049811963548814, "einstein": 0.0055859254602901405, "sagan": 0.6456451977498373, "darwin": 0.01949679559123049}, "confidence": 0.5221606058182879, "choice": "sagan"}}, "usage": {"input_tokens": 351, "output_tokens": 1}, "latency_ms": 171.33962500156485, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 351, "compiled_prompt_sha256": "a7a04181e9aebfc106fb6edbfdd1814d937a130ef09de73af59f3cc5ec826651", "compiled_ids_sha256": "d994e60a3d465bf24d7e166b1d1657d7fbe5a5c9d247dfe92152ecb34d2e61be", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.875, 18.5, 14.5, 19.25, 15.75], "runtime_probabilities": [0.016163410618901253, 0.008127473294734955, 0.30498120188713074, 0.005585925187915564, 0.6456451416015625, 0.019496794790029526], "decide_seconds": 0.16708731799735688, "probabilities_f64_minus_runtime_max_abs": 5.614827480737716e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:55Z"} {"id": "c-ibn_sina-01-dessert-03", "uid": "c059", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "prompt_sha256_by_order": ["5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de"], "request_sha256": "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "compiled_prompt_sha256_by_order": ["ba26f55ee8938b9cb0c3d4aabab1559ed19fcddd55502d02f38ac6dfdc9310ac", "f73da195f3d4ff8ad2b788d1861c94edc583a4558283eea237bcf2009ffd6d7a", "786941d7d59b0639240e0e5dfed6a5f5228060470d42e0be5ef919db367cebb4", "3313925cad3e57296a8ab4edabadbdda056a31d1195be93d79db72de90d06c2f", "307b482d6dcdd9895913213838827eb440c78b88ba5d7775b66b7bda68288dfe", "84a25c35093b3dbfe4158880bb0ee9aa7a46ad8c80652e1837ee591bd4a4f445"], "compiled_tokens_by_order": [393, 393, 393, 393, 393, 393], "choice": "ibn_sina", "probs": {"darwin": 0.05104336077228904, "einstein": 0.005726906360358921, "hypatia": 0.2937342604925405, "ibn_sina": 0.548767844724532, "sagan": 0.08415627463329346, "socrates": 0.016571353016986038}, "confidence": 0.35995122595143414, "seconds": 0.17709488983261204, "seconds_all_orders": 1.0625693389956723, "seconds_by_order": [0.1771149409978534, 0.17673977399681462, 0.17779691599935177, 0.17701775200112024, 0.17676299500089954, 0.17713696099963272], "argmax_stable": false, "spread": {"darwin": 0.137455883384852, "einstein": 0.01579893445061724, "hypatia": 0.8709310591259011, "ibn_sina": 0.9011734853509997, "sagan": 0.1304529623097178, "socrates": 0.013842072724861817}, "spread_2dp": {"darwin": 0.13, "einstein": 0.02, "hypatia": 0.87, "ibn_sina": 0.9, "sagan": 0.13, "socrates": 0.02}, "choices_by_order": ["ibn_sina", "ibn_sina", "hypatia", "ibn_sina", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.05104336077228904, "einstein": 0.005726906360358921, "hypatia": 0.2937342604925405, "ibn_sina": 0.548767844724532, "sagan": 0.08415627463329346, "socrates": 0.016571353016986038}, {"ibn_sina": 0.9163672108750788, "hypatia": 0.010836468616778509, "sagan": 0.021550891176140008, "socrates": 0.00398651401901235, "darwin": 0.04026236531888533, "einstein": 0.006996549994105141}, {"hypatia": 0.4144011507667432, "darwin": 0.006698170757747964, "sagan": 0.13453635607256267, "ibn_sina": 0.4144011507667432, "einstein": 0.018207515854702046, "socrates": 0.011755655781500856}, {"ibn_sina": 0.9128193028517687, "einstein": 0.002408581404084806, "darwin": 0.006969461382067319, "sagan": 0.008948965555198113, "hypatia": 0.06612440851475693, "socrates": 0.002729280292124222}, {"einstein": 0.009795552413227904, "sagan": 0.004083393762844871, "socrates": 0.004083393762844871, "ibn_sina": 0.04390062018828716, "darwin": 0.056369512130115686, "hypatia": 0.8817675277426796}, {"socrates": 0.015193725524079044, "ibn_sina": 0.015193725524079044, "hypatia": 0.7320746928444678, "einstein": 0.00594994840541225, "sagan": 0.08743385355936184, "darwin": 0.14415405414259996}], "prompt_tokens": 393, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.05104336077228904, "einstein": 0.005726906360358921, "hypatia": 0.2937342604925405, "ibn_sina": 0.548767844724532, "sagan": 0.08415627463329346, "socrates": 0.016571353016986038}, "confidence": 0.35995122595143414, "choice": "ibn_sina"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 177.08589200265124, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "ba26f55ee8938b9cb0c3d4aabab1559ed19fcddd55502d02f38ac6dfdc9310ac", "compiled_ids_sha256": "8f0bad5ac64183cef3d192f03fe419ee35d679774c7b48af85a668fb1670c955", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 14.6875, 18.625, 19.25, 17.375, 15.75], "runtime_probabilities": [0.05104335770010948, 0.005726906470954418, 0.29373428225517273, 0.5487678647041321, 0.08415627479553223, 0.016571354120969772], "decide_seconds": 0.17234384800030966, "probabilities_f64_minus_runtime_max_abs": 2.1762632251842717e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9163672108750788, "hypatia": 0.010836468616778509, "sagan": 0.021550891176140008, "socrates": 0.00398651401901235, "darwin": 0.04026236531888533, "einstein": 0.006996549994105141}, "confidence": 0.7779581855391203, "choice": "ibn_sina"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 176.71438400066108, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "f73da195f3d4ff8ad2b788d1861c94edc583a4558283eea237bcf2009ffd6d7a", "compiled_ids_sha256": "7b0b3941c0fe9f2cafbd86699092f7ac1408b7805ac1da098aac84e343d31302", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 15.1875, 15.875, 14.1875, 16.5, 14.75], "runtime_probabilities": [0.91636723279953, 0.010836468078196049, 0.021550891920924187, 0.003986514173448086, 0.0402623675763607, 0.006996549665927887], "decide_seconds": 0.17221699899891973, "probabilities_f64_minus_runtime_max_abs": 2.1924451254484723e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.4144011507667432, "darwin": 0.006698170757747964, "sagan": 0.13453635607256267, "ibn_sina": 0.4144011507667432, "einstein": 0.018207515854702046, "socrates": 0.011755655781500856}, "confidence": 0.35332716432188005, "choice": "hypatia"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 177.77146600201377, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "786941d7d59b0639240e0e5dfed6a5f5228060470d42e0be5ef919db367cebb4", "compiled_ids_sha256": "70b73c50c8d1f27c82fc4e875de53af42b6930960831645f28dfc49eba7593ba", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 15.0, 18.0, 19.125, 16.0, 15.5625], "runtime_probabilities": [0.4144011437892914, 0.006698170676827431, 0.13453635573387146, 0.4144011437892914, 0.01820751652121544, 0.011755656450986862], "decide_seconds": 0.1732635209991713, "probabilities_f64_minus_runtime_max_abs": 6.977451838530158e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9128193028517687, "einstein": 0.002408581404084806, "darwin": 0.006969461382067319, "sagan": 0.008948965555198113, "hypatia": 0.06612440851475693, "socrates": 0.002729280292124222}, "confidence": 0.7933183017997523, "choice": "ibn_sina"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 176.99237300257664, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "3313925cad3e57296a8ab4edabadbdda056a31d1195be93d79db72de90d06c2f", "compiled_ids_sha256": "752b3313009471cb89ab4fc593cb304d6d06b617638281f8ff8de33b64971599", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 13.8125, 14.875, 15.125, 17.125, 13.9375], "runtime_probabilities": [0.9128192663192749, 0.00240858132019639, 0.0069694602862000465, 0.00894896499812603, 0.06612440943717957, 0.0027292801532894373], "decide_seconds": 0.1724165070008894, "probabilities_f64_minus_runtime_max_abs": 3.65324938078615e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.009795552413227904, "sagan": 0.004083393762844871, "socrates": 0.004083393762844871, "ibn_sina": 0.04390062018828716, "darwin": 0.056369512130115686, "hypatia": 0.8817675277426796}, "confidence": 0.7206537328063385, "choice": "hypatia"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 176.73824500161572, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "307b482d6dcdd9895913213838827eb440c78b88ba5d7775b66b7bda68288dfe", "compiled_ids_sha256": "e44ae453b8a6f00b59c6a14c2f07974bdb8015a2f67b29c167404aba28aa9c8c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.125, 14.125, 16.5, 16.75, 19.5], "runtime_probabilities": [0.009795553050935268, 0.004083394072949886, 0.004083394072949886, 0.04390062019228935, 0.05636950954794884, 0.8817675113677979], "decide_seconds": 0.17223474899947178, "probabilities_f64_minus_runtime_max_abs": 1.6374881783853823e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015193725524079044, "ibn_sina": 0.015193725524079044, "hypatia": 0.7320746928444678, "einstein": 0.00594994840541225, "sagan": 0.08743385355936184, "darwin": 0.14415405414259996}, "confidence": 0.509808363112539, "choice": "hypatia"}}, "usage": {"input_tokens": 393, "output_tokens": 1}, "latency_ms": 177.11159200189286, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 393, "compiled_prompt_sha256": "84a25c35093b3dbfe4158880bb0ee9aa7a46ad8c80652e1837ee591bd4a4f445", "compiled_ids_sha256": "609a4b39d35c3b9de92d40c6499ecf122e9e9f6d389fecf4cc5b20e58e2b5e98", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 15.75, 19.625, 14.8125, 17.5, 18.0], "runtime_probabilities": [0.015193725936114788, 0.015193725936114788, 0.7320747375488281, 0.0059499493800103664, 0.08743385970592499, 0.14415405690670013], "decide_seconds": 0.17263941599958343, "probabilities_f64_minus_runtime_max_abs": 4.4704360324665515e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:56Z"} {"id": "c-ibn_sina-01-main-02", "uid": "c060", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "prompt_sha256_by_order": ["1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3"], "request_sha256": "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "compiled_prompt_sha256_by_order": ["ca53ab37565565807844480c303f0ac73c53fed29285230b4137aee8eb216a3b", "39e98b0948b727153d60ea5539b099ddb23e1bf077549d483ccddb71bd1d69b1", "ee11c68407ba0ef55c0c9352ac37e4063aca3c082e92b4c62f5e4899f6ee8bae", "75b52dfb0ca8e8c86961c20509cb18423a814d5ba413449941bfde6d18d78b60", "5aae23cbb9249a5311d01d124a3ea27e3e8e8c45865e504e191d918851f97541", "dd62f54b1437e983be0f8cf41ea0eaab16198c22cff47431dd9de6e3abfa4ab2"], "compiled_tokens_by_order": [317, 317, 317, 317, 317, 317], "choice": "hypatia", "probs": {"darwin": 0.011005895633587922, "einstein": 0.014131849726939526, "hypatia": 0.6396563608048016, "ibn_sina": 0.09809448000336796, "sagan": 0.11115560827021113, "socrates": 0.12595580556109187}, "confidence": 0.37014929790192297, "seconds": 0.16748356883302526, "seconds_all_orders": 1.0049014129981515, "seconds_by_order": [0.1692072419973556, 0.1702896740025608, 0.16572632899988093, 0.16755424499933724, 0.16537533199880272, 0.16674859100021422], "argmax_stable": true, "spread": {"darwin": 0.01352752657456048, "einstein": 0.04266916079643269, "hypatia": 0.266782877433875, "ibn_sina": 0.13332618078808423, "sagan": 0.28893270722425995, "socrates": 0.25553646997586593}, "spread_2dp": {"darwin": 0.01, "einstein": 0.049999999999999996, "hypatia": 0.2700000000000001, "ibn_sina": 0.13, "sagan": 0.29000000000000004, "socrates": 0.25}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.011005895633587922, "einstein": 0.014131849726939526, "hypatia": 0.6396563608048016, "ibn_sina": 0.09809448000336796, "sagan": 0.11115560827021113, "socrates": 0.12595580556109187}, {"ibn_sina": 0.03185404394414535, "hypatia": 0.564626438893653, "sagan": 0.14275992059054463, "socrates": 0.23537131768109212, "darwin": 0.010341493961808085, "einstein": 0.015046784928756734}, {"hypatia": 0.41155690340470924, "darwin": 0.006652197780904323, "sagan": 0.09183075776688464, "ibn_sina": 0.1514033236347737, "einstein": 0.05569817009025958, "socrates": 0.28285864732246846}, {"ibn_sina": 0.04547571083141358, "einstein": 0.013029009293826894, "darwin": 0.020179724355464803, "sagan": 0.3807634649911446, "hypatia": 0.4314615313390802, "socrates": 0.10909055918906999}, {"einstein": 0.02321151087624428, "sagan": 0.13357302260257148, "socrates": 0.13357302260257148, "ibn_sina": 0.018077142846689476, "darwin": 0.013225520233339009, "hypatia": 0.6783397808385843}, {"socrates": 0.027322177346602528, "ibn_sina": 0.027322177346602528, "hypatia": 0.6218498901354059, "einstein": 0.017640523542635452, "sagan": 0.29374108892564044, "darwin": 0.012124142703113088}], "prompt_tokens": 317, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.011005895633587922, "einstein": 0.014131849726939526, "hypatia": 0.6396563608048016, "ibn_sina": 0.09809448000336796, "sagan": 0.11115560827021113, "socrates": 0.12595580556109187}, "confidence": 0.37014929790192297, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 169.17627200018615, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "ca53ab37565565807844480c303f0ac73c53fed29285230b4137aee8eb216a3b", "compiled_ids_sha256": "46e60b1b41b20e85a4c648d7194e57c8c2da9d2e1ec5c2d9640a9a1b77e6ad6e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 15.6875, 19.5, 17.625, 17.75, 17.875], "runtime_probabilities": [0.011005896143615246, 0.014131849631667137, 0.6396563649177551, 0.09809448570013046, 0.11115560680627823, 0.12595580518245697], "decide_seconds": 0.16499943499729852, "probabilities_f64_minus_runtime_max_abs": 5.6967625061687954e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.03185404394414535, "hypatia": 0.564626438893653, "sagan": 0.14275992059054463, "socrates": 0.23537131768109212, "darwin": 0.010341493961808085, "einstein": 0.015046784928756734}, "confidence": 0.35185124014360647, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 170.26355400230386, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "39e98b0948b727153d60ea5539b099ddb23e1bf077549d483ccddb71bd1d69b1", "compiled_ids_sha256": "ea70e9e66ceee653cfc500254e77e108320a0811b68c1bc20786f61d5376abfd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 19.0, 17.625, 18.125, 15.0, 15.375], "runtime_probabilities": [0.031854044646024704, 0.5646264553070068, 0.14275991916656494, 0.23537130653858185, 0.010341494344174862, 0.015046784654259682], "decide_seconds": 0.1663856740015035, "probabilities_f64_minus_runtime_max_abs": 1.6413353787214646e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.41155690340470924, "darwin": 0.006652197780904323, "sagan": 0.09183075776688464, "ibn_sina": 0.1514033236347737, "einstein": 0.05569817009025958, "socrates": 0.28285864732246846}, "confidence": 0.20644142306859625, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 165.7010689996241, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "ee11c68407ba0ef55c0c9352ac37e4063aca3c082e92b4c62f5e4899f6ee8bae", "compiled_ids_sha256": "cedb33e032eafb968adad0ac88f71c2d1e7bc3d7be8a80e51d0c9a209b8eaa4c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 15.0, 17.625, 18.125, 17.125, 18.75], "runtime_probabilities": [0.4115568995475769, 0.006652197800576687, 0.09183076024055481, 0.15140333771705627, 0.05569817125797272, 0.282858669757843], "decide_seconds": 0.16184769899700768, "probabilities_f64_minus_runtime_max_abs": 2.243537455814959e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.04547571083141358, "einstein": 0.013029009293826894, "darwin": 0.020179724355464803, "sagan": 0.3807634649911446, "hypatia": 0.4314615313390802, "socrates": 0.10909055918906999}, "confidence": 0.30353700671258854, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 167.5295749992074, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "75b52dfb0ca8e8c86961c20509cb18423a814d5ba413449941bfde6d18d78b60", "compiled_ids_sha256": "f2b09fd1c84bd7c66b5e3210182bf1687c414d239b9c42cc230a6c4673b94ee6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 15.5, 15.9375, 18.875, 19.0, 17.625], "runtime_probabilities": [0.045475710183382034, 0.013029009103775024, 0.02017972618341446, 0.3807634711265564, 0.43146154284477234, 0.10909055918455124], "decide_seconds": 0.16365453499747673, "probabilities_f64_minus_runtime_max_abs": 1.1505692121271949e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.02321151087624428, "sagan": 0.13357302260257148, "socrates": 0.13357302260257148, "ibn_sina": 0.018077142846689476, "darwin": 0.013225520233339009, "hypatia": 0.6783397808385843}, "confidence": 0.4317522615134628, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 165.35039200243773, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "5aae23cbb9249a5311d01d124a3ea27e3e8e8c45865e504e191d918851f97541", "compiled_ids_sha256": "85e5b10bfeb721d4715841608bd82f139bbe59df9105f7ed4f8ec89dfccce547", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 17.75, 17.75, 15.75, 15.4375, 19.375], "runtime_probabilities": [0.023211508989334106, 0.1335730105638504, 0.1335730105638504, 0.018077142536640167, 0.013225520960986614, 0.6783397793769836], "decide_seconds": 0.16142726099860738, "probabilities_f64_minus_runtime_max_abs": 1.2038721075580128e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.027322177346602528, "ibn_sina": 0.027322177346602528, "hypatia": 0.6218498901354059, "einstein": 0.017640523542635452, "sagan": 0.29374108892564044, "darwin": 0.012124142703113088}, "confidence": 0.4548882433970738, "choice": "hypatia"}}, "usage": {"input_tokens": 317, "output_tokens": 1}, "latency_ms": 166.72125100012636, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 317, "compiled_prompt_sha256": "dd62f54b1437e983be0f8cf41ea0eaab16198c22cff47431dd9de6e3abfa4ab2", "compiled_ids_sha256": "db1b58832167ad7caebac5f82379980ec19eb64bd8a248074fe1a57eaa539002", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 16.25, 19.375, 15.8125, 18.625, 15.4375], "runtime_probabilities": [0.027322176843881607, 0.027322176843881607, 0.6218498945236206, 0.017640523612499237, 0.2937410771846771, 0.012124142609536648], "decide_seconds": 0.16291114100022241, "probabilities_f64_minus_runtime_max_abs": 1.17409633126897e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:57Z"} {"id": "c-ibn_sina-01-main-02", "uid": "c061", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "prompt_sha256_by_order": ["0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c"], "request_sha256": "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "compiled_prompt_sha256_by_order": ["45e5e8ec908f7291f32e80dc6bdc7bd7e15ac15096efb3d82a2feb0b9cc1ed9e", "43ebe18dbb17a2f7624a1dd7f6cca575d005e584f01fb9b1e2f20ea142d0b939", "066fbd33d05df3931f35e871f7ba2a9e5622956e46f7f969d1fe1edfe24c21cd", "fb91358cbe69e57c48bca479dd8f25387d2ac5b7735a3ec39700029a75f69438", "5f495cfa45655914b778c03fdb85692182eca55025caf29f24329f6cbe089f84", "b7f7f1c571d02fc4949e129890bf2703743efafaeab498becfd9f3ee28d5edd1"], "compiled_tokens_by_order": [352, 352, 352, 352, 352, 352], "choice": "ibn_sina", "probs": {"darwin": 0.025332028014288237, "einstein": 0.19925183447701736, "hypatia": 0.12085234661429776, "ibn_sina": 0.37225143427264196, "sagan": 0.0830605221447373, "socrates": 0.19925183447701736}, "confidence": 0.12606570643062687, "seconds": 0.17107531749995056, "seconds_all_orders": 1.0264519049997034, "seconds_by_order": [0.17161766299977899, 0.17113603699908708, 0.17093473900240497, 0.17085308899913798, 0.17079554899828508, 0.1711148280010093], "argmax_stable": false, "spread": {"darwin": 0.06033059345707349, "einstein": 0.4486925244246016, "hypatia": 0.2993150729745548, "ibn_sina": 0.2338174831188198, "sagan": 0.10295698121879754, "socrates": 0.208388884391444}, "spread_2dp": {"darwin": 0.060000000000000005, "einstein": 0.44999999999999996, "hypatia": 0.3, "ibn_sina": 0.22999999999999998, "sagan": 0.1, "socrates": 0.21}, "choices_by_order": ["ibn_sina", "hypatia", "einstein", "einstein", "hypatia", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.025332028014288237, "einstein": 0.19925183447701736, "hypatia": 0.12085234661429776, "ibn_sina": 0.37225143427264196, "sagan": 0.0830605221447373, "socrates": 0.19925183447701736}, {"ibn_sina": 0.15157189149950048, "hypatia": 0.32087769682245704, "sagan": 0.03382026042428397, "socrates": 0.2498998015554784, "darwin": 0.049208188557484905, "einstein": 0.19462216114079525}, {"hypatia": 0.17775271181401694, "darwin": 0.03088878969093225, "sagan": 0.012876363255986706, "ibn_sina": 0.13843395115382215, "einstein": 0.4831819664833597, "socrates": 0.15686621760188219}, {"ibn_sina": 0.16288042062696273, "einstein": 0.4427548876020387, "darwin": 0.00810933863754103, "sagan": 0.014232332540684466, "hypatia": 0.20914259996581042, "socrates": 0.16288042062696273}, {"einstein": 0.03448944205875808, "sagan": 0.009881390590865777, "socrates": 0.12037998119657446, "ibn_sina": 0.37079644635412445, "darwin": 0.04428532021082455, "hypatia": 0.42016741958885256}, {"socrates": 0.04151091716403438, "ibn_sina": 0.18603902375376163, "hypatia": 0.14488733738126924, "einstein": 0.44628441779665684, "sagan": 0.11283837180966332, "darwin": 0.06843993209461452}], "prompt_tokens": 352, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.025332028014288237, "einstein": 0.19925183447701736, "hypatia": 0.12085234661429776, "ibn_sina": 0.37225143427264196, "sagan": 0.0830605221447373, "socrates": 0.19925183447701736}, "confidence": 0.12606570643062687, "choice": "ibn_sina"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 171.588992998295, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "45e5e8ec908f7291f32e80dc6bdc7bd7e15ac15096efb3d82a2feb0b9cc1ed9e", "compiled_ids_sha256": "8cb23d0f97d0241542c05f0a7bceae8684c28142b6f39ba5f8ef5db574533fac", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 18.0, 17.5, 18.625, 17.125, 18.0], "runtime_probabilities": [0.025332028046250343, 0.19925184547901154, 0.12085234373807907, 0.3722514510154724, 0.08306052535772324, 0.19925184547901154], "decide_seconds": 0.16723579699828406, "probabilities_f64_minus_runtime_max_abs": 1.674283045538516e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.15157189149950048, "hypatia": 0.32087769682245704, "sagan": 0.03382026042428397, "socrates": 0.2498998015554784, "darwin": 0.049208188557484905, "einstein": 0.19462216114079525}, "confidence": 0.11901058628077932, "choice": "hypatia"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 171.1106269976881, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "43ebe18dbb17a2f7624a1dd7f6cca575d005e584f01fb9b1e2f20ea142d0b939", "compiled_ids_sha256": "a8ee1a7e9630195985a411143fd0c6e80e1295719119b12ea1fe94ee441b41bc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.625, 18.375, 16.125, 18.125, 16.5, 17.875], "runtime_probabilities": [0.15157189965248108, 0.32087770104408264, 0.0338202603161335, 0.24989980459213257, 0.04920819029211998, 0.19462215900421143], "decide_seconds": 0.16683159999956843, "probabilities_f64_minus_runtime_max_abs": 8.152980596198844e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.17775271181401694, "darwin": 0.03088878969093225, "sagan": 0.012876363255986706, "ibn_sina": 0.13843395115382215, "einstein": 0.4831819664833597, "socrates": 0.15686621760188219}, "confidence": 0.22631769950339198, "choice": "einstein"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 170.90947900214815, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "066fbd33d05df3931f35e871f7ba2a9e5622956e46f7f969d1fe1edfe24c21cd", "compiled_ids_sha256": "0b5fcc8216b374c3e0a4862bbe036d81d4ce232c3505647362254d74cc3e8ea7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.25, 16.5, 15.625, 18.0, 19.25, 18.125], "runtime_probabilities": [0.17775273323059082, 0.030888788402080536, 0.012876362539827824, 0.13843394815921783, 0.4831819534301758, 0.15686620771884918], "decide_seconds": 0.16670053099733195, "probabilities_f64_minus_runtime_max_abs": 2.141657387544349e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.16288042062696273, "einstein": 0.4427548876020387, "darwin": 0.00810933863754103, "sagan": 0.014232332540684466, "hypatia": 0.20914259996581042, "socrates": 0.16288042062696273}, "confidence": 0.23052297155171997, "choice": "einstein"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 170.8289700000023, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "fb91358cbe69e57c48bca479dd8f25387d2ac5b7735a3ec39700029a75f69438", "compiled_ids_sha256": "2f5846c01b3a8524e3d9fc66159650b093a31d98ac143b8b91d700661e92f2dd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.0, 19.0, 15.0, 15.5625, 18.25, 18.0], "runtime_probabilities": [0.16288042068481445, 0.442754864692688, 0.008109338581562042, 0.01423233188688755, 0.20914259552955627, 0.16288042068481445], "decide_seconds": 0.16663168099694303, "probabilities_f64_minus_runtime_max_abs": 2.2909350694266806e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03448944205875808, "sagan": 0.009881390590865777, "socrates": 0.12037998119657446, "ibn_sina": 0.37079644635412445, "darwin": 0.04428532021082455, "hypatia": 0.42016741958885256}, "confidence": 0.28179681827243475, "choice": "hypatia"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 170.77158000029158, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "5f495cfa45655914b778c03fdb85692182eca55025caf29f24329f6cbe089f84", "compiled_ids_sha256": "5faa349e6b2578dcfc39544f5f485b1ba45f81024969b1902bafc7b968d831b4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 15.25, 17.75, 18.875, 16.75, 19.0], "runtime_probabilities": [0.03448944166302681, 0.009881391189992428, 0.1203799843788147, 0.37079644203186035, 0.04428531974554062, 0.4201674163341522], "decide_seconds": 0.16666425099901971, "probabilities_f64_minus_runtime_max_abs": 4.322264102185613e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.04151091716403438, "ibn_sina": 0.18603902375376163, "hypatia": 0.14488733738126924, "einstein": 0.44628441779665684, "sagan": 0.11283837180966332, "darwin": 0.06843993209461452}, "confidence": 0.15465925495094512, "choice": "einstein"}}, "usage": {"input_tokens": 352, "output_tokens": 1}, "latency_ms": 171.09032799999113, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 352, "compiled_prompt_sha256": "b7f7f1c571d02fc4949e129890bf2703743efafaeab498becfd9f3ee28d5edd1", "compiled_ids_sha256": "70dfa8a1bfc1a14d9b1dae26fb2a6a808bd5e8ba25f6e508b3b99bda4d4530df", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 17.875, 17.625, 18.75, 17.375, 16.875], "runtime_probabilities": [0.04151091352105141, 0.1860390156507492, 0.14488732814788818, 0.4462844133377075, 0.11283836513757706, 0.06843993067741394], "decide_seconds": 0.16697386899977573, "probabilities_f64_minus_runtime_max_abs": 9.2333810597367e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:58Z"} {"id": "c-ibn_sina-01-main-02", "uid": "c062", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "prompt_sha256_by_order": ["c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de"], "request_sha256": "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "compiled_prompt_sha256_by_order": ["2c19043fdf98431a6068bfeda9e535e1c58513f3ad737c128fbdbff9a34cc6d6", "f8a96c99bb0bcb4ffb9a654fd9d37a52f317b95761ec49f9d334f4d4ea2c2559", "6a1aa8de9a9f2ff697666631a887e22076082309700ab0e3c41d489443820ec7", "3c84328cb34dea419c7c25ce54449f742b7fbab8e99d65bb1e2c8b8b33532292", "c86433d742b975ae4f4c0e95443299abc7f3ec1b30de25134bfbc69e36d8cbc4", "248a411ff2de31f890240ce1d14cd50f09d8b969ac7249c4ad6ef1227d5c37a9"], "compiled_tokens_by_order": [365, 365, 365, 365, 365, 365], "choice": "ibn_sina", "probs": {"darwin": 0.01240644622192247, "einstein": 0.019215479806888494, "hypatia": 0.20658587678064214, "ibn_sina": 0.3859532291580259, "sagan": 0.1105774515174546, "socrates": 0.2652615165150664}, "confidence": 0.20796079853233174, "seconds": 0.17110785733287534, "seconds_all_orders": 1.026647143997252, "seconds_by_order": [0.17183556199961458, 0.17114166700048372, 0.171031657999265, 0.17097194799862336, 0.17079816999830655, 0.17086813900095876], "argmax_stable": false, "spread": {"darwin": 0.06854320581717101, "einstein": 0.03801001056234331, "hypatia": 0.43909848146922653, "ibn_sina": 0.7021072710733676, "sagan": 0.14621619783430212, "socrates": 0.26379111967633334}, "spread_2dp": {"darwin": 0.07, "einstein": 0.04, "hypatia": 0.43, "ibn_sina": 0.7, "sagan": 0.15, "socrates": 0.27}, "choices_by_order": ["ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "hypatia", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01240644622192247, "einstein": 0.019215479806888494, "hypatia": 0.20658587678064214, "ibn_sina": 0.3859532291580259, "sagan": 0.1105774515174546, "socrates": 0.2652615165150664}, {"ibn_sina": 0.9128503045741966, "hypatia": 0.03539504742881063, "sagan": 0.015706457086726962, "socrates": 0.010794879563497927, "darwin": 0.012232201078420372, "einstein": 0.013021110268347551}, {"hypatia": 0.09600252088441569, "darwin": 0.014722463407259119, "sagan": 0.0101185912572016, "ibn_sina": 0.8038192659670094, "einstein": 0.040019804750105706, "socrates": 0.035317353734008565}, {"ibn_sina": 0.9780221545381967, "einstein": 0.0020097941877623995, "darwin": 0.0012190017945901445, "sagan": 0.001470396838733093, "hypatia": 0.015808255801984472, "socrates": 0.001470396838733093}, {"einstein": 0.01877624345799279, "sagan": 0.14768659467303522, "socrates": 0.03295333352117059, "ibn_sina": 0.27591488346482923, "darwin": 0.06976220761176115, "hypatia": 0.454906737271211}, {"socrates": 0.012108860031526846, "ibn_sina": 0.8488966255585173, "hypatia": 0.0422640743339309, "einstein": 0.017618287386685045, "sagan": 0.06149386530265493, "darwin": 0.017618287386685045}], "prompt_tokens": 365, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01240644622192247, "einstein": 0.019215479806888494, "hypatia": 0.20658587678064214, "ibn_sina": 0.3859532291580259, "sagan": 0.1105774515174546, "socrates": 0.2652615165150664}, "confidence": 0.20796079853233174, "choice": "ibn_sina"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 171.80723199999193, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "2c19043fdf98431a6068bfeda9e535e1c58513f3ad737c128fbdbff9a34cc6d6", "compiled_ids_sha256": "3d9a463164b741f3095477f08090e3b5559a38fc08b00bb779a1cc9e5a019438", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 15.875, 18.25, 18.875, 17.625, 18.5], "runtime_probabilities": [0.012406446970999241, 0.019215479493141174, 0.20658589899539948, 0.38595324754714966, 0.11057745665311813, 0.26526153087615967], "decide_seconds": 0.16730499700133805, "probabilities_f64_minus_runtime_max_abs": 2.2214757339211744e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9128503045741966, "hypatia": 0.03539504742881063, "sagan": 0.015706457086726962, "socrates": 0.010794879563497927, "darwin": 0.012232201078420372, "einstein": 0.013021110268347551}, "confidence": 0.7622347591851156, "choice": "ibn_sina"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 171.11581700009992, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "f8a96c99bb0bcb4ffb9a654fd9d37a52f317b95761ec49f9d334f4d4ea2c2559", "compiled_ids_sha256": "ad58a7c79fd06d1f6f83b57b67c4e720c704b32d0d1e5314b865e5c53d7f6575", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 16.375, 15.5625, 15.1875, 15.3125, 15.375], "runtime_probabilities": [0.9128503203392029, 0.035395048558712006, 0.015706457197666168, 0.01079487893730402, 0.012232202105224133, 0.01302111055701971], "decide_seconds": 0.1667369310016511, "probabilities_f64_minus_runtime_max_abs": 1.576500630040556e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.09600252088441569, "darwin": 0.014722463407259119, "sagan": 0.0101185912572016, "ibn_sina": 0.8038192659670094, "einstein": 0.040019804750105706, "socrates": 0.035317353734008565}, "confidence": 0.5780845400527612, "choice": "ibn_sina"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 170.9991980023915, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "6a1aa8de9a9f2ff697666631a887e22076082309700ab0e3c41d489443820ec7", "compiled_ids_sha256": "7c1ce254745608e98c221fb7b287465928144449e665abbbcaec4193030f7365", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 15.625, 15.25, 19.625, 16.625, 16.5], "runtime_probabilities": [0.09600251913070679, 0.014722463674843311, 0.010118591599166393, 0.8038192391395569, 0.04001980274915695, 0.035317353904247284], "decide_seconds": 0.16676607000044896, "probabilities_f64_minus_runtime_max_abs": 2.68274524817258e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9780221545381967, "einstein": 0.0020097941877623995, "darwin": 0.0012190017945901445, "sagan": 0.001470396838733093, "hypatia": 0.015808255801984472, "socrates": 0.001470396838733093}, "confidence": 0.9290446912043153, "choice": "ibn_sina"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 170.94779900071444, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "3c84328cb34dea419c7c25ce54449f742b7fbab8e99d65bb1e2c8b8b33532292", "compiled_ids_sha256": "694b9761221e04b9caad2dd622fb93a70d59f49fd89be0956251e67b53042bd4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.875, 13.6875, 13.1875, 13.375, 15.75, 13.375], "runtime_probabilities": [0.9780220985412598, 0.0020097941160202026, 0.0012190018314868212, 0.0014703967608511448, 0.01580825448036194, 0.0014703967608511448], "decide_seconds": 0.16664344200034975, "probabilities_f64_minus_runtime_max_abs": 5.5996936976221434e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.01877624345799279, "sagan": 0.14768659467303522, "socrates": 0.03295333352117059, "ibn_sina": 0.27591488346482923, "darwin": 0.06976220761176115, "hypatia": 0.454906737271211}, "confidence": 0.23598917482592396, "choice": "hypatia"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 170.77463999885367, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "c86433d742b975ae4f4c0e95443299abc7f3ec1b30de25134bfbc69e36d8cbc4", "compiled_ids_sha256": "607de14276f46f3b59e16d3b9f5efdf6ae84efd14a524bad73f359d95a73868c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 18.0, 16.5, 18.625, 17.25, 19.125], "runtime_probabilities": [0.018776245415210724, 0.14768660068511963, 0.03295333683490753, 0.27591490745544434, 0.0697622150182724, 0.45490676164627075], "decide_seconds": 0.16659636199983652, "probabilities_f64_minus_runtime_max_abs": 2.4375059748393824e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.012108860031526846, "ibn_sina": 0.8488966255585173, "hypatia": 0.0422640743339309, "einstein": 0.017618287386685045, "sagan": 0.06149386530265493, "darwin": 0.017618287386685045}, "confidence": 0.6427891839749595, "choice": "ibn_sina"}}, "usage": {"input_tokens": 365, "output_tokens": 1}, "latency_ms": 170.84443000203464, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 365, "compiled_prompt_sha256": "248a411ff2de31f890240ce1d14cd50f09d8b969ac7249c4ad6ef1227d5c37a9", "compiled_ids_sha256": "7291562e49e886c24323a5132473b11430f918e9dae4dd1eb1321f1528c7426c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 19.375, 16.375, 15.5, 16.75, 15.5], "runtime_probabilities": [0.012108859606087208, 0.8488966226577759, 0.042264074087142944, 0.017618287354707718, 0.06149386614561081, 0.017618287354707718], "decide_seconds": 0.16662670200093999, "probabilities_f64_minus_runtime_max_abs": 2.9007414248383157e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:48:59Z"} {"id": "c-ibn_sina-01-main-03", "uid": "c063", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "prompt_sha256_by_order": ["9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf"], "request_sha256": "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "compiled_prompt_sha256_by_order": ["7b9355b2e2fdcf6d623298737a222d53b12dc9d6c71de4bf927676f9c9b5c339", "839704d4054c24973da64c8cfcf8e745b6d22abc74db8054480aee813f8306a9", "6a8aba645314a1fa85e7ee3fb5056567e43519e1f258201c67d2c8e5ef9f4acb", "b7eee888397e052b348ea35ecf9cfeb4181bf95fe56f71b61bb5c95fc6063f01", "da835f51ed0a9ecd870385d5d74d76bcb254737df1ae2bf3e36c28195a5a1b74", "a3ac943ebad6c98f5b9a2b2a20b87f37c802031c79fe24ffe29857101c8d108c"], "compiled_tokens_by_order": [390, 390, 390, 390, 390, 390], "choice": "ibn_sina", "probs": {"darwin": 0.005358042328016558, "einstein": 0.005033414954046281, "hypatia": 0.013682240404678024, "ibn_sina": 0.9591991053964662, "sagan": 0.006071457375458539, "socrates": 0.01065573954133446}, "confidence": 0.8701198527892271, "seconds": 0.17737047250011528, "seconds_all_orders": 1.0642228350006917, "seconds_by_order": [0.17739944899949478, 0.17733307000162313, 0.17684842300150194, 0.17684216300040134, 0.1786202500006766, 0.1771794799969939], "argmax_stable": true, "spread": {"darwin": 0.005343735354016519, "einstein": 0.01000389356113654, "hypatia": 0.015860560630562768, "ibn_sina": 0.07402654234268558, "sagan": 0.005240862843015651, "socrates": 0.045594646539840265}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.01, "ibn_sina": 0.06999999999999995, "sagan": 0.01, "socrates": 0.04}, "choices_by_order": ["ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.005358042328016558, "einstein": 0.005033414954046281, "hypatia": 0.013682240404678024, "ibn_sina": 0.9591991053964662, "sagan": 0.006071457375458539, "socrates": 0.01065573954133446}, {"ibn_sina": 0.971394101929972, "hypatia": 0.005426163020841361, "sagan": 0.002261961275972766, "socrates": 0.015701123309521354, "darwin": 0.003091734493465469, "einstein": 0.002124915970227054}, {"hypatia": 0.007375942618527078, "darwin": 0.005744389887198637, "sagan": 0.0028884606314604063, "ibn_sina": 0.9660578385931722, "einstein": 0.007851650046718162, "socrates": 0.010081718222923612}, {"ibn_sina": 0.9791610642367164, "einstein": 0.0031164550319425894, "darwin": 0.0015670520013871716, "sagan": 0.001668118172312786, "hypatia": 0.009017761640622754, "socrates": 0.005469548917018198}, {"einstein": 0.012128809531363593, "sagan": 0.003474962110939264, "socrates": 0.05106419545685846, "ibn_sina": 0.9051345218940309, "darwin": 0.006910787355403691, "hypatia": 0.02128672365140413}, {"socrates": 0.010700839129228699, "ibn_sina": 0.9632588409215413, "hypatia": 0.01005250806106302, "einstein": 0.0036981110478752828, "sagan": 0.006908981015328437, "darwin": 0.0053807198249632565}], "prompt_tokens": 390, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.005358042328016558, "einstein": 0.005033414954046281, "hypatia": 0.013682240404678024, "ibn_sina": 0.9591991053964662, "sagan": 0.006071457375458539, "socrates": 0.01065573954133446}, "confidence": 0.8701198527892271, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 177.3660100006964, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "7b9355b2e2fdcf6d623298737a222d53b12dc9d6c71de4bf927676f9c9b5c339", "compiled_ids_sha256": "45337ce0418c787e52b602dcd036092f44cf3be9bf63859bfa2bb7606fe58827", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.4375, 14.375, 15.375, 19.625, 14.5625, 15.125], "runtime_probabilities": [0.005358042661100626, 0.005033415276557207, 0.01368224062025547, 0.9591991305351257, 0.006071457639336586, 0.010655740275979042], "decide_seconds": 0.17270166500020423, "probabilities_f64_minus_runtime_max_abs": 2.5138659487389248e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.971394101929972, "hypatia": 0.005426163020841361, "sagan": 0.002261961275972766, "socrates": 0.015701123309521354, "darwin": 0.003091734493465469, "einstein": 0.002124915970227054}, "confidence": 0.9071057796083908, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 177.30551999920863, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "839704d4054c24973da64c8cfcf8e745b6d22abc74db8054480aee813f8306a9", "compiled_ids_sha256": "b5aa16ab2d17d63b1810c22bd88461f8d6f7e39e05c2a0a0b4eadacac0e39e4b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 14.4375, 13.5625, 15.5, 13.875, 13.5], "runtime_probabilities": [0.9713941216468811, 0.0054261633194983006, 0.002261961344629526, 0.01570112258195877, 0.0030917346011847258, 0.00212491606362164], "decide_seconds": 0.17258952599877375, "probabilities_f64_minus_runtime_max_abs": 1.9716909105405023e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.007375942618527078, "darwin": 0.005744389887198637, "sagan": 0.0028884606314604063, "ibn_sina": 0.9660578385931722, "einstein": 0.007851650046718162, "socrates": 0.010081718222923612}, "confidence": 0.8880974947893874, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 176.8215029987914, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "6a8aba645314a1fa85e7ee3fb5056567e43519e1f258201c67d2c8e5ef9f4acb", "compiled_ids_sha256": "848f0fa51b134d4a7cf5177855ec47ac62d13963b35d46a3885bcdd595f8984d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 14.625, 13.9375, 19.75, 14.9375, 15.1875], "runtime_probabilities": [0.0073759425431489944, 0.005744389723986387, 0.0028884608764201403, 0.9660578966140747, 0.007851650938391685, 0.010081718675792217], "decide_seconds": 0.17229801900248276, "probabilities_f64_minus_runtime_max_abs": 5.802090252693404e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9791610642367164, "einstein": 0.0031164550319425894, "darwin": 0.0015670520013871716, "sagan": 0.001668118172312786, "hypatia": 0.009017761640622754, "socrates": 0.005469548917018198}, "confidence": 0.9272531035432507, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 176.81501300103264, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "b7eee888397e052b348ea35ecf9cfeb4181bf95fe56f71b61bb5c95fc6063f01", "compiled_ids_sha256": "6ea46cdfe48e2dfc197a3a837b76cb45f1f557976675947c87105bc0f7a710c1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 13.875, 13.1875, 13.25, 14.9375, 14.4375], "runtime_probabilities": [0.9791612029075623, 0.003116455627605319, 0.001567052211612463, 0.001668118522502482, 0.00901776272803545, 0.005469549912959337], "decide_seconds": 0.1721941190007783, "probabilities_f64_minus_runtime_max_abs": 1.3867084580621025e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.012128809531363593, "sagan": 0.003474962110939264, "socrates": 0.05106419545685846, "ibn_sina": 0.9051345218940309, "darwin": 0.006910787355403691, "hypatia": 0.02128672365140413}, "confidence": 0.759101853457738, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 178.59157000202686, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "da835f51ed0a9ecd870385d5d74d76bcb254737df1ae2bf3e36c28195a5a1b74", "compiled_ids_sha256": "91546ae5eebe1f1aed50cb1c711035ba6e2d70be59095a20191a73e8344d8d5e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.3125, 17.0, 19.875, 15.0, 16.125], "runtime_probabilities": [0.01212881039828062, 0.003474962431937456, 0.051064200699329376, 0.9051345586776733, 0.006910787429660559, 0.021286725997924805], "decide_seconds": 0.1740224950008269, "probabilities_f64_minus_runtime_max_abs": 3.678364246528787e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.010700839129228699, "ibn_sina": 0.9632588409215413, "hypatia": 0.01005250806106302, "einstein": 0.0036981110478752828, "sagan": 0.006908981015328437, "darwin": 0.0053807198249632565}, "confidence": 0.8805376582950862, "choice": "ibn_sina"}}, "usage": {"input_tokens": 390, "output_tokens": 1}, "latency_ms": 177.1487209989573, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 390, "compiled_prompt_sha256": "a3ac943ebad6c98f5b9a2b2a20b87f37c802031c79fe24ffe29857101c8d108c", "compiled_ids_sha256": "1774314ec5ff3f00a168eac60e6d655dd13b8a4c0f459b9de303e2d58363da99", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 19.625, 15.0625, 14.0625, 14.6875, 14.4375], "runtime_probabilities": [0.010700841434299946, 0.9632589221000671, 0.010052509605884552, 0.0036981115117669106, 0.006908981129527092, 0.0053807208314538], "decide_seconds": 0.17242048700063606, "probabilities_f64_minus_runtime_max_abs": 8.117852579037077e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:00Z"} {"id": "c-ibn_sina-01-main-03", "uid": "c064", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "prompt_sha256_by_order": ["eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032"], "request_sha256": "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "compiled_prompt_sha256_by_order": ["0dc80c8adce05b741ce8a4c3f17b9cc1f78bc491ba12ae7dfad74cd0ca49ecbb", "92f29f32e14c0d979e049d00ae719f31c0d0da0122d96aa375125d516c967b4e", "1fdc75835ec0479b5bbe1e83c16b140a074ebf5531ff54502deb61b080c6ec67", "2fb060e55afe946d016fe92b383c97bff1d9af578c8faffb60ecf93e3304d0b5", "46b4e4f2c74a7952af24aff3e6d1e10460bdd83b6948a8b801d2554d33b0880c", "141aa7fb13c0d156a5285ca3b26ef84e029202c3c89660e85c97a44d86fff8f5"], "compiled_tokens_by_order": [371, 371, 371, 371, 371, 371], "choice": "socrates", "probs": {"darwin": 0.009503313115928181, "einstein": 0.007878523292290173, "hypatia": 0.19087914648258913, "ibn_sina": 0.02011851402429541, "sagan": 0.01667883394066164, "socrates": 0.7549416691442354}, "confidence": 0.577167551396031, "seconds": 0.17185724833385999, "seconds_all_orders": 1.03114349000316, "seconds_by_order": [0.17475187899981393, 0.17161933399984264, 0.1714540749999287, 0.17109435700331233, 0.17103812799905427, 0.1711857170012081], "argmax_stable": false, "spread": {"darwin": 0.01230210054433134, "einstein": 0.023517978512842375, "hypatia": 0.830491372978131, "ibn_sina": 0.014984263009611344, "sagan": 0.03597861162436221, "socrates": 0.8475235071220755}, "spread_2dp": {"darwin": 0.02, "einstein": 0.03, "hypatia": 0.83, "ibn_sina": 0.01, "sagan": 0.04, "socrates": 0.8500000000000001}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.009503313115928181, "einstein": 0.007878523292290173, "hypatia": 0.19087914648258913, "ibn_sina": 0.02011851402429541, "sagan": 0.01667883394066164, "socrates": 0.7549416691442354}, {"ibn_sina": 0.024463652936702378, "hypatia": 0.10963848598696276, "sagan": 0.019052312063890884, "socrates": 0.8101249235594897, "darwin": 0.008999674971366185, "einstein": 0.02772095048158802}, {"hypatia": 0.0314215957814595, "darwin": 0.007463283167298652, "sagan": 0.01394324860647766, "ibn_sina": 0.013098469878980907, "einstein": 0.015799770589156343, "socrates": 0.918273631976627}, {"ibn_sina": 0.01216170264223, "einstein": 0.004202971968745646, "darwin": 0.0034843861449861437, "sagan": 0.01216170264223, "hypatia": 0.21557133737262743, "socrates": 0.7524178992291808}, {"einstein": 0.015628871806055523, "sagan": 0.04814031426659221, "socrates": 0.5175571534619925, "ibn_sina": 0.009479389927091034, "darwin": 0.00612035413786173, "hypatia": 0.403073916400407}, {"socrates": 0.07075012485455152, "ibn_sina": 0.0115496403880628, "hypatia": 0.8619129687595906, "einstein": 0.006580786724936776, "sagan": 0.033419992583540885, "darwin": 0.015786486689317484}], "prompt_tokens": 371, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.009503313115928181, "einstein": 0.007878523292290173, "hypatia": 0.19087914648258913, "ibn_sina": 0.02011851402429541, "sagan": 0.01667883394066164, "socrates": 0.7549416691442354}, "confidence": 0.577167551396031, "choice": "socrates"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 174.72095899938722, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "0dc80c8adce05b741ce8a4c3f17b9cc1f78bc491ba12ae7dfad74cd0ca49ecbb", "compiled_ids_sha256": "5948b85d5511809552d45e91339724803cfce0ca4a60973ed39e61cc07e354a8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 15.1875, 18.375, 16.125, 15.9375, 19.75], "runtime_probabilities": [0.009503312408924103, 0.00787852331995964, 0.19087913632392883, 0.020118512213230133, 0.016678832471370697, 0.7549416422843933], "decide_seconds": 0.17022998399988865, "probabilities_f64_minus_runtime_max_abs": 2.6859842128246214e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.024463652936702378, "hypatia": 0.10963848598696276, "sagan": 0.019052312063890884, "socrates": 0.8101249235594897, "darwin": 0.008999674971366185, "einstein": 0.02772095048158802}, "confidence": 0.597619266586157, "choice": "socrates"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.5921340000932, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "92f29f32e14c0d979e049d00ae719f31c0d0da0122d96aa375125d516c967b4e", "compiled_ids_sha256": "349e335094b5b2613ba976f92a3d52b0bda3da0dc84aae7ab7bc6348372f525f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 17.625, 15.875, 19.625, 15.125, 16.25], "runtime_probabilities": [0.024463653564453125, 0.10963848978281021, 0.01905231364071369, 0.810124933719635, 0.008999675512313843, 0.027720950543880463], "decide_seconds": 0.1672444070027268, "probabilities_f64_minus_runtime_max_abs": 1.0160145280657673e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0314215957814595, "darwin": 0.007463283167298652, "sagan": 0.01394324860647766, "ibn_sina": 0.013098469878980907, "einstein": 0.015799770589156343, "socrates": 0.918273631976627}, "confidence": 0.7737044395519468, "choice": "socrates"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.42616500132135, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "1fdc75835ec0479b5bbe1e83c16b140a074ebf5531ff54502deb61b080c6ec67", "compiled_ids_sha256": "7bbfab7f9a1495424321586c307489982b246392ed3e26f8727e7a1dd8c34555", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 15.0625, 15.6875, 15.625, 15.8125, 19.875], "runtime_probabilities": [0.03142159432172775, 0.007463283371180296, 0.013943247497081757, 0.013098468072712421, 0.015799768269062042, 0.9182735681533813], "decide_seconds": 0.166960670001572, "probabilities_f64_minus_runtime_max_abs": 6.382324568843245e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01216170264223, "einstein": 0.004202971968745646, "darwin": 0.0034843861449861437, "sagan": 0.01216170264223, "hypatia": 0.21557133737262743, "socrates": 0.7524178992291808}, "confidence": 0.6122285857060666, "choice": "socrates"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.0678569979791, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "2fb060e55afe946d016fe92b383c97bff1d9af578c8faffb60ecf93e3304d0b5", "compiled_ids_sha256": "256699a7033ee5bfb677c22b959892945d72206216e42a670785a4afa3eb83d5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 14.6875, 14.5, 15.75, 18.625, 19.875], "runtime_probabilities": [0.01216170284897089, 0.004202972166240215, 0.00348438648506999, 0.01216170284897089, 0.2155713438987732, 0.7524179220199585], "decide_seconds": 0.16669654100041953, "probabilities_f64_minus_runtime_max_abs": 2.2790777709502663e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.015628871806055523, "sagan": 0.04814031426659221, "socrates": 0.5175571534619925, "ibn_sina": 0.009479389927091034, "darwin": 0.00612035413786173, "hypatia": 0.403073916400407}, "confidence": 0.4455088735230801, "choice": "socrates"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.01172800175846, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "46b4e4f2c74a7952af24aff3e6d1e10460bdd83b6948a8b801d2554d33b0880c", "compiled_ids_sha256": "5d706c18d28ff7f3ad2f27cce23ac56a9bd9f104f4fa89d73ea315c1e28ef8de", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.875, 17.0, 19.375, 15.375, 14.9375, 19.125], "runtime_probabilities": [0.01562887243926525, 0.04814031347632408, 0.5175571441650391, 0.00947939045727253, 0.006120353937149048, 0.4030739367008209], "decide_seconds": 0.16671382099957555, "probabilities_f64_minus_runtime_max_abs": 2.0300413905438575e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.07075012485455152, "ibn_sina": 0.0115496403880628, "hypatia": 0.8619129687595906, "einstein": 0.006580786724936776, "sagan": 0.033419992583540885, "darwin": 0.015786486689317484}, "confidence": 0.6767834176821996, "choice": "hypatia"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.15914699752466, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "141aa7fb13c0d156a5285ca3b26ef84e029202c3c89660e85c97a44d86fff8f5", "compiled_ids_sha256": "708e0f0b5dfb29dc2304483dcab07e249867406883d1f8fa3310ccc684659188", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 15.6875, 20.0, 15.125, 16.75, 16.0], "runtime_probabilities": [0.07075012475252151, 0.011549641378223896, 0.8619130253791809, 0.006580786779522896, 0.03341999277472496, 0.015786487609148026], "decide_seconds": 0.16675534099704237, "probabilities_f64_minus_runtime_max_abs": 5.6619590349527016e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:01Z"} {"id": "c-ibn_sina-01-main-03", "uid": "c065", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "prompt_sha256_by_order": ["c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086"], "request_sha256": "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "compiled_prompt_sha256_by_order": ["667abdaec59327957c3b1ce1c6570cf662d99312a7e39336fd34c6a5b9174833", "1457bcd521f631b20130859f96550f0e527fce3c361cf06075a56425819cbf27", "3c56b6d22318a0264796bbd2a4497b08d207ec80950a4da84c215ed194a553a0", "99f47c3be7c674e23eb09726b53e8cfc0e27222cd22465dcfd6d5b65babaa6e2", "295b7162c7889b2500fdcaebdf11e9fdfbd4accabd136238e467217b468723b1", "504ae730d37539aad2381411fd9a1e632d32a31cba4b065b59b4f7bc3fff7132"], "compiled_tokens_by_order": [395, 395, 395, 395, 395, 395], "choice": "ibn_sina", "probs": {"darwin": 0.004810256728357337, "einstein": 0.0024187489811652555, "hypatia": 0.008986742686964743, "ibn_sina": 0.9757929832332813, "sagan": 0.0042450366634120594, "socrates": 0.0037462317068191883}, "confidence": 0.9159385147255001, "seconds": 0.17914850933326912, "seconds_all_orders": 1.0748910559996148, "seconds_by_order": [0.17735023999921395, 0.17779050700119114, 0.17745730900060153, 0.1770167520007817, 0.18706900499819312, 0.1782072429996333], "argmax_stable": true, "spread": {"darwin": 0.008034436118293981, "einstein": 0.0071092763216763985, "hypatia": 0.010620906509638102, "ibn_sina": 0.023985365378751844, "sagan": 0.002676775760516477, "socrates": 0.0054838124527019794}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.01, "ibn_sina": 0.020000000000000018, "sagan": 0.0, "socrates": 0.01}, "choices_by_order": ["ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.004810256728357337, "einstein": 0.0024187489811652555, "hypatia": 0.008986742686964743, "ibn_sina": 0.9757929832332813, "sagan": 0.0042450366634120594, "socrates": 0.0037462317068191883}, {"ibn_sina": 0.9559310721873019, "hypatia": 0.016447702638585675, "sagan": 0.0028581821482073244, "socrates": 0.008803820810184935, "darwin": 0.010619434971726993, "einstein": 0.005339787243993279}, {"hypatia": 0.01005860221089731, "darwin": 0.00887668529544744, "sagan": 0.002244377522186867, "ibn_sina": 0.9638427990412586, "einstein": 0.00887668529544744, "socrates": 0.006100850634762498}, {"ibn_sina": 0.9796790954466158, "einstein": 0.00275171695578846, "darwin": 0.002584998853433011, "sagan": 0.002428381690272828, "hypatia": 0.009022532543126845, "socrates": 0.0035332745107630008}, {"einstein": 0.0045379133880218, "sagan": 0.0015682609028955822, "socrates": 0.0033200083574829553, "ibn_sina": 0.9799164375660537, "darwin": 0.004830583656598377, "hypatia": 0.005826796128947573}, {"socrates": 0.003982917605625838, "ibn_sina": 0.9745878049697685, "hypatia": 0.006990238092042789, "einstein": 0.001767408973771041, "sagan": 0.004239793721515093, "darwin": 0.008431836637276814}], "prompt_tokens": 395, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.004810256728357337, "einstein": 0.0024187489811652555, "hypatia": 0.008986742686964743, "ibn_sina": 0.9757929832332813, "sagan": 0.0042450366634120594, "socrates": 0.0037462317068191883}, "confidence": 0.9159385147255001, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 177.32088999764528, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "667abdaec59327957c3b1ce1c6570cf662d99312a7e39336fd34c6a5b9174833", "compiled_ids_sha256": "5cf614b764b9c0e5e91de93bb9bc44413b47819378f9850fb3926ac12e4ae96b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 13.875, 15.1875, 19.875, 14.4375, 14.3125], "runtime_probabilities": [0.004810256417840719, 0.0024187490344047546, 0.008986742235720158, 0.9757929444313049, 0.004245036747306585, 0.003746231785044074], "decide_seconds": 0.17259627600287786, "probabilities_f64_minus_runtime_max_abs": 3.880197640970806e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9559310721873019, "hypatia": 0.016447702638585675, "sagan": 0.0028581821482073244, "socrates": 0.008803820810184935, "darwin": 0.010619434971726993, "einstein": 0.005339787243993279}, "confidence": 0.8631194853890342, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 177.7639870015264, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "1457bcd521f631b20130859f96550f0e527fce3c361cf06075a56425819cbf27", "compiled_ids_sha256": "7d4bf7e48f3e8b271911fdf7932a4c21070a4290a0a62adbd27723fb12f0f704", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 15.6875, 13.9375, 15.0625, 15.25, 14.5625], "runtime_probabilities": [0.9559311270713806, 0.016447702422738075, 0.002858182415366173, 0.008803821168839931, 0.010619436390697956, 0.005339787807315588], "decide_seconds": 0.1731328220012074, "probabilities_f64_minus_runtime_max_abs": 5.488407872622503e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.01005860221089731, "darwin": 0.00887668529544744, "sagan": 0.002244377522186867, "ibn_sina": 0.9638427990412586, "einstein": 0.00887668529544744, "socrates": 0.006100850634762498}, "confidence": 0.882556523285681, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 177.43070900178282, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "3c56b6d22318a0264796bbd2a4497b08d207ec80950a4da84c215ed194a553a0", "compiled_ids_sha256": "bfbfd824b862da1bc19553b2be0dded328783033e6d2c26d9518916cd345f071", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 15.1875, 13.8125, 19.875, 15.1875, 14.8125], "runtime_probabilities": [0.010058602318167686, 0.008876685053110123, 0.002244377275928855, 0.9638427495956421, 0.008876685053110123, 0.0061008501797914505], "decide_seconds": 0.17261477599822683, "probabilities_f64_minus_runtime_max_abs": 4.944561649011803e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9796790954466158, "einstein": 0.00275171695578846, "darwin": 0.002584998853433011, "sagan": 0.002428381690272828, "hypatia": 0.009022532543126845, "socrates": 0.0035332745107630008}, "confidence": 0.9281248040555062, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.99132199777523, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "99f47c3be7c674e23eb09726b53e8cfc0e27222cd22465dcfd6d5b65babaa6e2", "compiled_ids_sha256": "0c282b5357c92272a6afe1f967802a734487fae6388a7a4b826c8f1f78b42064", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 13.875, 13.8125, 13.75, 15.0625, 14.125], "runtime_probabilities": [0.9796792268753052, 0.0027517173439264297, 0.00258499919436872, 0.00242838216945529, 0.00902253296226263, 0.0035332748666405678], "decide_seconds": 0.17240201700042235, "probabilities_f64_minus_runtime_max_abs": 1.3142868937610075e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0045379133880218, "sagan": 0.0015682609028955822, "socrates": 0.0033200083574829553, "ibn_sina": 0.9799164375660537, "darwin": 0.004830583656598377, "hypatia": 0.005826796128947573}, "confidence": 0.9279019877644917, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 187.04230499861296, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "295b7162c7889b2500fdcaebdf11e9fdfbd4accabd136238e467217b468723b1", "compiled_ids_sha256": "766600d72a832559ac0695e171dde43c726438684c7c5d244cb26104ae981f62", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 13.4375, 14.1875, 19.875, 14.5625, 14.75], "runtime_probabilities": [0.004537913948297501, 0.001568260951898992, 0.0033200085163116455, 0.9799164533615112, 0.004830583464354277, 0.005826795939356089], "decide_seconds": 0.18243332999918493, "probabilities_f64_minus_runtime_max_abs": 1.579545749752498e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.003982917605625838, "ibn_sina": 0.9745878049697685, "hypatia": 0.006990238092042789, "einstein": 0.001767408973771041, "sagan": 0.004239793721515093, "darwin": 0.008431836637276814}, "confidence": 0.9126987281834543, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 178.17896400083555, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "504ae730d37539aad2381411fd9a1e632d32a31cba4b065b59b4f7bc3fff7132", "compiled_ids_sha256": "3576abe73dfbb8680d649f45c0bb1da7d0d2142b03e8f68de99b13f53e1b767b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 19.875, 14.9375, 13.5625, 14.4375, 15.125], "runtime_probabilities": [0.003982917405664921, 0.9745877385139465, 0.006990237161517143, 0.0017674088012427092, 0.0042397934012115, 0.008431836031377316], "decide_seconds": 0.173141242001293, "probabilities_f64_minus_runtime_max_abs": 6.645582195652366e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:02Z"} {"id": "c-ibn_sina-01-starter-01", "uid": "c066", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "prompt_sha256_by_order": ["43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423"], "request_sha256": "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "compiled_prompt_sha256_by_order": ["0cc652ed158d4456d057ae4ea6f240eebcd0b64c04d9666f8ff066084455a03d", "051c8a009024080fc05ff9992a9126e6e5556581499cdf44658c8923a02ebecc", "6d83fc1bf8b4240715f520976bd2d427b922117c4f560bbf0823743856b3302e", "467fe3acfa3d4644b06eab8263530d26f099342c4ce4b607700c9b598b3b629c", "a77dd33c62eb50ff5d0c72ef241a7c36c6e40315ebbcdcddd425295a9513da25", "bd644fac9a14c031262ff3cac101ea851c45ce1bcebbbb073935e6c94ab052ef"], "compiled_tokens_by_order": [293, 293, 293, 293, 293, 293], "choice": "ibn_sina", "probs": {"darwin": 0.0034581336005359673, "einstein": 0.002232738863376828, "hypatia": 0.02895459813753609, "ibn_sina": 0.9588446036068184, "sagan": 0.0030517921911967054, "socrates": 0.0034581336005359673}, "confidence": 0.8809242301361283, "seconds": 0.16712142300032914, "seconds_all_orders": 1.0027285380019748, "seconds_by_order": [0.16903943299985258, 0.168205099998886, 0.17037102300309925, 0.16509549400143442, 0.16490719499779516, 0.1651102930009074], "argmax_stable": false, "spread": {"darwin": 0.0038301100995425997, "einstein": 0.0034196514707997717, "hypatia": 0.485451564132664, "ibn_sina": 0.5582246816140305, "sagan": 0.0031388310807099135, "socrates": 0.06615917006896051}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.48, "ibn_sina": 0.5599999999999999, "sagan": 0.01, "socrates": 0.07}, "choices_by_order": ["ibn_sina", "hypatia", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0034581336005359673, "einstein": 0.002232738863376828, "hypatia": 0.02895459813753609, "ibn_sina": 0.9588446036068184, "sagan": 0.0030517921911967054, "socrates": 0.0034581336005359673}, {"ibn_sina": 0.40061992199278795, "hypatia": 0.5144061622702001, "sagan": 0.005714536275910162, "socrates": 0.06961730366949648, "darwin": 0.005714536275910162, "einstein": 0.0039275395156951436}, {"hypatia": 0.21944803567540863, "darwin": 0.003775812022881547, "sagan": 0.002940605360150574, "ibn_sina": 0.7659489058484975, "einstein": 0.00333214241493464, "socrates": 0.004554498678127145}, {"ibn_sina": 0.6533274926441934, "einstein": 0.0056523903341766, "darwin": 0.003649457490056382, "sagan": 0.003428348038341446, "hypatia": 0.3086100555112683, "socrates": 0.02533225598196379}, {"einstein": 0.005267873489775077, "sagan": 0.0026488531394148894, "socrates": 0.008685255073949483, "ibn_sina": 0.8859201024734352, "darwin": 0.004102623998957924, "hypatia": 0.09337529182446745}, {"socrates": 0.008990094488565064, "ibn_sina": 0.9170145681091314, "hypatia": 0.0664283125106945, "einstein": 0.0031068935200413116, "sagan": 0.0025757051952002485, "darwin": 0.0018844261763675623}], "prompt_tokens": 293, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0034581336005359673, "einstein": 0.002232738863376828, "hypatia": 0.02895459813753609, "ibn_sina": 0.9588446036068184, "sagan": 0.0030517921911967054, "socrates": 0.0034581336005359673}, "confidence": 0.8809242301361283, "choice": "ibn_sina"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 169.0124630003993, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "0cc652ed158d4456d057ae4ea6f240eebcd0b64c04d9666f8ff066084455a03d", "compiled_ids_sha256": "d87679564b3455f5e970ac7dffd32d0e50699c17f522bec77e89352be1475266", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 13.8125, 16.375, 19.875, 14.125, 14.25], "runtime_probabilities": [0.0034581334330141544, 0.002232738770544529, 0.02895459719002247, 0.9588445425033569, 0.0030517922714352608, 0.0034581334330141544], "decide_seconds": 0.16521369299880462, "probabilities_f64_minus_runtime_max_abs": 6.110346151277923e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.40061992199278795, "hypatia": 0.5144061622702001, "sagan": 0.005714536275910162, "socrates": 0.06961730366949648, "darwin": 0.005714536275910162, "einstein": 0.0039275395156951436}, "confidence": 0.45600438407066246, "choice": "hypatia"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 168.18082999816397, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "051c8a009024080fc05ff9992a9126e6e5556581499cdf44658c8923a02ebecc", "compiled_ids_sha256": "42010f20150d52a892158a49a35b8dc42661bbd9cef267f488ea8e85c38b6c78", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 19.375, 14.875, 17.375, 14.875, 14.5], "runtime_probabilities": [0.40061989426612854, 0.514406144618988, 0.005714536178857088, 0.06961730122566223, 0.005714536178857088, 0.003927539102733135], "decide_seconds": 0.16439445899959537, "probabilities_f64_minus_runtime_max_abs": 2.772665941286334e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.21944803567540863, "darwin": 0.003775812022881547, "sagan": 0.002940605360150574, "ibn_sina": 0.7659489058484975, "einstein": 0.00333214241493464, "socrates": 0.004554498678127145}, "confidence": 0.6546264618493756, "choice": "ibn_sina"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 170.34647399850655, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "6d83fc1bf8b4240715f520976bd2d427b922117c4f560bbf0823743856b3302e", "compiled_ids_sha256": "fa864540899bcf7e47fb7c7d78cb1c292d4d5b4501554ccc6e6a0b7a78281c3f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 14.4375, 14.1875, 19.75, 14.3125, 14.625], "runtime_probabilities": [0.2194480448961258, 0.0037758119869977236, 0.0029406053945422173, 0.7659488916397095, 0.003332142485305667, 0.004554498475044966], "decide_seconds": 0.16661886200017761, "probabilities_f64_minus_runtime_max_abs": 1.4208788035041664e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.6533274926441934, "einstein": 0.0056523903341766, "darwin": 0.003649457490056382, "sagan": 0.003428348038341446, "hypatia": 0.3086100555112683, "socrates": 0.02533225598196379}, "confidence": 0.5517013389999523, "choice": "ibn_sina"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 165.07266400003573, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "467fe3acfa3d4644b06eab8263530d26f099342c4ce4b607700c9b598b3b629c", "compiled_ids_sha256": "e5a64b699cf87fc6cc342b8d82c5b11e0deb42f8f8c463dc91054d882f080894", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 14.75, 14.3125, 14.25, 18.75, 16.25], "runtime_probabilities": [0.653327465057373, 0.005652390420436859, 0.0036494575906544924, 0.003428348107263446, 0.30861005187034607, 0.02533225528895855], "decide_seconds": 0.1614512910018675, "probabilities_f64_minus_runtime_max_abs": 2.758682038361826e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.005267873489775077, "sagan": 0.0026488531394148894, "socrates": 0.008685255073949483, "ibn_sina": 0.8859201024734352, "darwin": 0.004102623998957924, "hypatia": 0.09337529182446745}, "confidence": 0.7567541048069786, "choice": "ibn_sina"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 164.88517499965383, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "a77dd33c62eb50ff5d0c72ef241a7c36c6e40315ebbcdcddd425295a9513da25", "compiled_ids_sha256": "9664907577861e4b904f89b63d1f5468ba3f85330b6520d9a18880b5d9189f8d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 13.9375, 15.125, 19.75, 14.375, 17.5], "runtime_probabilities": [0.005267873872071505, 0.002648853464052081, 0.008685256354510784, 0.8859201669692993, 0.004102624487131834, 0.0933752954006195], "decide_seconds": 0.1613566819978587, "probabilities_f64_minus_runtime_max_abs": 6.449586409384978e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.008990094488565064, "ibn_sina": 0.9170145681091314, "hypatia": 0.0664283125106945, "einstein": 0.0031068935200413116, "sagan": 0.0025757051952002485, "darwin": 0.0018844261763675623}, "confidence": 0.8063088012984377, "choice": "ibn_sina"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 165.08761400109506, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "bd644fac9a14c031262ff3cac101ea851c45ce1bcebbbb073935e6c94ab052ef", "compiled_ids_sha256": "6c7024adcc488c3914596d4009087d1e4d32391d116887aeb732f953b654898e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 19.75, 17.125, 14.0625, 13.875, 13.5625], "runtime_probabilities": [0.008990094996988773, 0.9170145392417908, 0.06642831116914749, 0.0031068935059010983, 0.0025757052935659885, 0.001884426106698811], "decide_seconds": 0.1615133409977716, "probabilities_f64_minus_runtime_max_abs": 2.8867340629012972e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:03Z"} {"id": "c-ibn_sina-01-starter-02", "uid": "c067", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "prompt_sha256_by_order": ["131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b"], "request_sha256": "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "compiled_prompt_sha256_by_order": ["ecd96e4320ca6e089fea50dd88dd138efc7728b40394e645c167cf30bfe00de9", "e837597389e921bfabccb4624e12765961ff0f28a20786da6f360d2a2be2fa5a", "0653d2d3d0724423c604902ed6aebd51e2efdef2221e90d3a635410da984e7ab", "0526b088281e8c62edb7029cfe9a42f3dcd5ad11943bc20565928378bc4ae0fb", "64ed332f520e76af3430affbbf35fc200b98cb08d6293555684649e2bd84d334", "f7579a22982616d24a0f7349bdac6063ce8b01e3024ef360e84120e360461e58"], "compiled_tokens_by_order": [311, 311, 311, 311, 311, 311], "choice": "sagan", "probs": {"darwin": 0.030636360596756142, "einstein": 0.023859621623211615, "hypatia": 0.030636360596756142, "ibn_sina": 0.014471592043620354, "sagan": 0.4501980325698279, "socrates": 0.4501980325698279}, "confidence": 0.39580512138388513, "seconds": 0.1687492521656774, "seconds_all_orders": 1.0124955129940645, "seconds_by_order": [0.1692121819978638, 0.1680446509999456, 0.1657048790002591, 0.16544810999766923, 0.17185913199864444, 0.1722265589996823], "argmax_stable": false, "spread": {"darwin": 0.23332399450795838, "einstein": 0.14691730039660217, "hypatia": 0.07233816825382888, "ibn_sina": 0.05010113804747494, "sagan": 0.5448706444556402, "socrates": 0.5737375148367387}, "spread_2dp": {"darwin": 0.24000000000000002, "einstein": 0.15, "hypatia": 0.06999999999999999, "ibn_sina": 0.049999999999999996, "sagan": 0.55, "socrates": 0.5800000000000001}, "choices_by_order": ["sagan", "socrates", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.030636360596756142, "einstein": 0.023859621623211615, "hypatia": 0.030636360596756142, "ibn_sina": 0.014471592043620354, "sagan": 0.4501980325698279, "socrates": 0.4501980325698279}, {"ibn_sina": 0.026072183923544302, "hypatia": 0.021614599647173143, "sagan": 0.15971148934888899, "socrates": 0.5574478720746366, "darwin": 0.07544236565686799, "einstein": 0.15971148934888899}, {"hypatia": 0.052933839421935185, "darwin": 0.02206610721076575, "sagan": 0.34517199221503886, "ibn_sina": 0.036380860319938965, "einstein": 0.041224915592779125, "socrates": 0.5022222852395422}, {"ibn_sina": 0.02708529822453688, "einstein": 0.012794188952286812, "darwin": 0.02708529822453688, "sagan": 0.18800923420571128, "hypatia": 0.08880927384522991, "socrates": 0.6562167065476983}, {"einstein": 0.04438014509387299, "sagan": 0.06457273009109529, "socrates": 0.47713152510421036, "ibn_sina": 0.06457273009109529, "darwin": 0.2553901017187241, "hypatia": 0.09395276790100203}, {"socrates": 0.08247919171095949, "ibn_sina": 0.038960411461371205, "hypatia": 0.038960411461371205, "einstein": 0.032299315557766134, "sagan": 0.6094433745467355, "darwin": 0.19785729526179655}], "prompt_tokens": 311, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.030636360596756142, "einstein": 0.023859621623211615, "hypatia": 0.030636360596756142, "ibn_sina": 0.014471592043620354, "sagan": 0.4501980325698279, "socrates": 0.4501980325698279}, "confidence": 0.39580512138388513, "choice": "sagan"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 169.1713920008624, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "ecd96e4320ca6e089fea50dd88dd138efc7728b40394e645c167cf30bfe00de9", "compiled_ids_sha256": "ea289ecc97718490e97f10028ee7f3325d7a73f38345e7ea80efb30da1bcf2a9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 15.6875, 15.9375, 15.1875, 18.625, 18.625], "runtime_probabilities": [0.030636359006166458, 0.023859620094299316, 0.030636359006166458, 0.014471592381596565, 0.4501980245113373, 0.4501980245113373], "decide_seconds": 0.16525453299982473, "probabilities_f64_minus_runtime_max_abs": 8.058490597040446e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.026072183923544302, "hypatia": 0.021614599647173143, "sagan": 0.15971148934888899, "socrates": 0.5574478720746366, "darwin": 0.07544236565686799, "einstein": 0.15971148934888899}, "confidence": 0.28302695094265795, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 168.0191610030306, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "e837597389e921bfabccb4624e12765961ff0f28a20786da6f360d2a2be2fa5a", "compiled_ids_sha256": "5a5e18daf95230b049598d927906f664779fdb31a02062ba877faf9076e9b3d6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 15.75, 17.75, 19.0, 17.0, 17.75], "runtime_probabilities": [0.02607218362390995, 0.021614598110318184, 0.15971149504184723, 0.5574478507041931, 0.0754423588514328, 0.15971149504184723], "decide_seconds": 0.16405262199987192, "probabilities_f64_minus_runtime_max_abs": 2.1370443525903227e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.052933839421935185, "darwin": 0.02206610721076575, "sagan": 0.34517199221503886, "ibn_sina": 0.036380860319938965, "einstein": 0.041224915592779125, "socrates": 0.5022222852395422}, "confidence": 0.3276040970668903, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 165.68183900017175, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "0653d2d3d0724423c604902ed6aebd51e2efdef2221e90d3a635410da984e7ab", "compiled_ids_sha256": "35556c5e1ad3b0e67bb75b007d39f6a764d42dbd19732cd55ad31e2c38eede96", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 15.625, 18.375, 16.125, 16.25, 18.75], "runtime_probabilities": [0.05293383449316025, 0.022066107019782066, 0.3451719880104065, 0.03638086095452309, 0.04122491553425789, 0.5022222399711609], "decide_seconds": 0.16192371799843386, "probabilities_f64_minus_runtime_max_abs": 4.526838126572841e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02708529822453688, "einstein": 0.012794188952286812, "darwin": 0.02708529822453688, "sagan": 0.18800923420571128, "hypatia": 0.08880927384522991, "socrates": 0.6562167065476983}, "confidence": 0.4101104511941155, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 165.42399100217153, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "0526b088281e8c62edb7029cfe9a42f3dcd5ad11943bc20565928378bc4ae0fb", "compiled_ids_sha256": "ed53035ba24f6c1d397aff1bf0030dd16703f34b3d1bbf11757e2d1f4bc6102f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 15.0625, 15.8125, 17.75, 17.0, 19.0], "runtime_probabilities": [0.027085300534963608, 0.01279419008642435, 0.027085300534963608, 0.18800924718379974, 0.08880927413702011, 0.6562167406082153], "decide_seconds": 0.16158478099896456, "probabilities_f64_minus_runtime_max_abs": 3.406051707344204e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.04438014509387299, "sagan": 0.06457273009109529, "socrates": 0.47713152510421036, "ibn_sina": 0.06457273009109529, "darwin": 0.2553901017187241, "hypatia": 0.09395276790100203}, "confidence": 0.20974386192806072, "choice": "socrates"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 171.83589200067217, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "64ed332f520e76af3430affbbf35fc200b98cb08d6293555684649e2bd84d334", "compiled_ids_sha256": "3c389e9fc6ff42665e40d1f441700c2ecc2a8bb6feb63c26baef50b274e2a6c7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 16.625, 18.625, 16.625, 18.0, 17.0], "runtime_probabilities": [0.04438014328479767, 0.06457272917032242, 0.47713154554367065, 0.06457272917032242, 0.25539010763168335, 0.09395276755094528], "decide_seconds": 0.16797456100175623, "probabilities_f64_minus_runtime_max_abs": 2.0439460290955225e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.08247919171095949, "ibn_sina": 0.038960411461371205, "hypatia": 0.038960411461371205, "einstein": 0.032299315557766134, "sagan": 0.6094433745467355, "darwin": 0.19785729526179655}, "confidence": 0.3347775732791216, "choice": "sagan"}}, "usage": {"input_tokens": 311, "output_tokens": 1}, "latency_ms": 172.20280900073703, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 311, "compiled_prompt_sha256": "f7579a22982616d24a0f7349bdac6063ce8b01e3024ef360e84120e360461e58", "compiled_ids_sha256": "19c30f7bea095ef8843df96eaa7a7b5c0e2aa3a462a81a3fc19304ab57357dc2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 16.0, 16.0, 15.8125, 18.75, 17.625], "runtime_probabilities": [0.08247918635606766, 0.03896040841937065, 0.03896040841937065, 0.03229931369423866, 0.6094433665275574, 0.19785729050636292], "decide_seconds": 0.16837429900260759, "probabilities_f64_minus_runtime_max_abs": 8.01917809933883e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:04Z"} {"id": "c-ibn_sina-01-starter-03", "uid": "c068", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "prompt_sha256_by_order": ["34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf"], "request_sha256": "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "compiled_prompt_sha256_by_order": ["0ad9f34f9bb88bf2b729ba7edfdf37192f81ae6be3b2bb68a023efff32a6297e", "b5975b47ab93a6abc1b31cba510da59f0acc0a5ef0f2ca35a1697e3811c3e734", "ac11a482d4195265088df28225a75a7c7fa4bfb7a2ddf6df2890362c4f4a918f", "8bcad8b709de4d85c01f2b966a8a44a7db39a90c37172f1044258555fad31378", "190de18c143d4fe6df134521b00c7220f889a90dafe55c70e1d39d0cf8de89d7", "2ff33004630d8056e520d0ea2e5bea927888a17c9cf79b258e1aa233284c5b44"], "compiled_tokens_by_order": [371, 371, 371, 371, 371, 371], "choice": "ibn_sina", "probs": {"darwin": 0.0012352855031388882, "einstein": 0.0010240876513682648, "hypatia": 0.0038049472199925918, "ibn_sina": 0.9910868011936753, "sagan": 0.0016884350938721952, "socrates": 0.0011604433379527885}, "confidence": 0.9642717007168818, "seconds": 0.17216623916707854, "seconds_all_orders": 1.0329974350024713, "seconds_by_order": [0.17164992299876758, 0.17468625000037719, 0.17154200400182162, 0.17133937499966123, 0.170921699002065, 0.1728581839997787], "argmax_stable": true, "spread": {"darwin": 0.0015106931737530222, "einstein": 0.0025302604710672246, "hypatia": 0.0028427154576238056, "ibn_sina": 0.005473315870541562, "sagan": 0.0012595253057696024, "socrates": 0.0018607314936971494}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.010000000000000009, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0012352855031388882, "einstein": 0.0010240876513682648, "hypatia": 0.0038049472199925918, "ibn_sina": 0.9910868011936753, "sagan": 0.0016884350938721952, "socrates": 0.0011604433379527885}, {"ibn_sina": 0.9917523084560569, "hypatia": 0.004888929615795539, "sagan": 0.0007497416391375136, "socrates": 0.0010247753184509866, "darwin": 0.0009626873206016964, "einstein": 0.0006215576499573009}, {"hypatia": 0.0035714778281869326, "darwin": 0.0009030110105043978, "sagan": 0.0007032656821029256, "ibn_sina": 0.9902718165451118, "einstein": 0.0031518181210245257, "socrates": 0.0013986108130694369}, {"ibn_sina": 0.9957451324156533, "einstein": 0.0006240600593349846, "darwin": 0.00040292345771213573, "sagan": 0.0004289097881025929, "hypatia": 0.0020462141581717333, "socrates": 0.0007527601210253105}, {"einstein": 0.001586443908518908, "sagan": 0.0008491622327390809, "socrates": 0.0007977140938829495, "ibn_sina": 0.9912779565975841, "darwin": 0.0019136166314651579, "hypatia": 0.003575106535809848}, {"socrates": 0.00261349161472246, "ibn_sina": 0.9904774610125242, "hypatia": 0.0029614739803258774, "einstein": 0.0012345260246639964, "sagan": 0.0013989012551187464, "darwin": 0.001314146112644717}], "prompt_tokens": 371, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0012352855031388882, "einstein": 0.0010240876513682648, "hypatia": 0.0038049472199925918, "ibn_sina": 0.9910868011936753, "sagan": 0.0016884350938721952, "socrates": 0.0011604433379527885}, "confidence": 0.9642717007168818, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.6218729998218, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "0ad9f34f9bb88bf2b729ba7edfdf37192f81ae6be3b2bb68a023efff32a6297e", "compiled_ids_sha256": "1fa8144179d5b9344965adbb89b17f9bf2ac697e24f6cd4b3cb376295fba968c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 13.0, 14.3125, 19.875, 13.5, 13.125], "runtime_probabilities": [0.0012352855410426855, 0.0010240876581519842, 0.0038049472495913506, 0.9910867810249329, 0.0016884352080523968, 0.0011604432947933674], "decide_seconds": 0.16715479800041066, "probabilities_f64_minus_runtime_max_abs": 2.0168742453918753e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9917523084560569, "hypatia": 0.004888929615795539, "sagan": 0.0007497416391375136, "socrates": 0.0010247753184509866, "darwin": 0.0009626873206016964, "einstein": 0.0006215576499573009}, "confidence": 0.9676568976968752, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 174.6603600004164, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "b5975b47ab93a6abc1b31cba510da59f0acc0a5ef0f2ca35a1697e3811c3e734", "compiled_ids_sha256": "be787a016be8e0db152762e222cfa3f08a65b376b275768e4c67f1b759dfa9e9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 14.4375, 12.5625, 12.875, 12.8125, 12.375], "runtime_probabilities": [0.9917522072792053, 0.004888928961008787, 0.0007497415645048022, 0.001024775207042694, 0.0009626871324144304, 0.000621557526756078], "decide_seconds": 0.17030246300055296, "probabilities_f64_minus_runtime_max_abs": 1.0117685156352252e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0035714778281869326, "darwin": 0.0009030110105043978, "sagan": 0.0007032656821029256, "ibn_sina": 0.9902718165451118, "einstein": 0.0031518181210245257, "socrates": 0.0013986108130694369}, "confidence": 0.9617211650007059, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.51610500150127, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "ac11a482d4195265088df28225a75a7c7fa4bfb7a2ddf6df2890362c4f4a918f", "compiled_ids_sha256": "6a98be94c1b57bc9c628b884f2a7055e00de5a20206781ad8e766b970e9d3f0c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 12.875, 12.625, 19.875, 14.125, 13.3125], "runtime_probabilities": [0.0035714777186512947, 0.0009030109504237771, 0.0007032656576484442, 0.9902717471122742, 0.0031518181785941124, 0.0013986107660457492], "decide_seconds": 0.16710862800027826, "probabilities_f64_minus_runtime_max_abs": 6.943283759053287e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.9957451324156533, "einstein": 0.0006240600593349846, "darwin": 0.00040292345771213573, "sagan": 0.0004289097881025929, "hypatia": 0.0020462141581717333, "socrates": 0.0007527601210253105}, "confidence": 0.981353695823399, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.31452600006014, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "8bcad8b709de4d85c01f2b966a8a44a7db39a90c37172f1044258555fad31378", "compiled_ids_sha256": "2424748256d78181298665b17bdfc3793da88705ecbf1f6b87a392bf2e91d9f0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.875, 12.5, 12.0625, 12.125, 13.6875, 12.6875], "runtime_probabilities": [0.9957452416419983, 0.0006240601069293916, 0.0004029234987683594, 0.00042890984332188964, 0.0020462144166231155, 0.0007527602137997746], "decide_seconds": 0.16683981000096537, "probabilities_f64_minus_runtime_max_abs": 1.0922634496868255e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.001586443908518908, "sagan": 0.0008491622327390809, "socrates": 0.0007977140938829495, "ibn_sina": 0.9912779565975841, "darwin": 0.0019136166314651579, "hypatia": 0.003575106535809848}, "confidence": 0.9649930494632796, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 170.89644899897394, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "190de18c143d4fe6df134521b00c7220f889a90dafe55c70e1d39d0cf8de89d7", "compiled_ids_sha256": "ba54ae93476e974e164bb01ac096bcc1ceace9c5633deb1b1e80d019a3a559cd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 12.6875, 12.625, 19.75, 13.5, 14.125], "runtime_probabilities": [0.001586443861015141, 0.0008491621701978147, 0.0007977140485309064, 0.9912779331207275, 0.001913616550154984, 0.0035751063842326403], "decide_seconds": 0.16657806199873448, "probabilities_f64_minus_runtime_max_abs": 2.3476856569892846e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.00261349161472246, "ibn_sina": 0.9904774610125242, "hypatia": 0.0029614739803258774, "einstein": 0.0012345260246639964, "sagan": 0.0013989012551187464, "darwin": 0.001314146112644717}, "confidence": 0.961801936205523, "choice": "ibn_sina"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 172.83368399876053, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "2ff33004630d8056e520d0ea2e5bea927888a17c9cf79b258e1aa233284c5b44", "compiled_ids_sha256": "9e59dfc8370018894c44a9c7a769c1711f03da0724329efe0f2480980e7df9bb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.9375, 19.875, 14.0625, 13.1875, 13.3125, 13.25], "runtime_probabilities": [0.0026134916115552187, 0.990477442741394, 0.0029614740051329136, 0.0012345260474830866, 0.0013989012222737074, 0.001314146094955504], "decide_seconds": 0.16851079700063565, "probabilities_f64_minus_runtime_max_abs": 1.8271130164393412e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:05Z"} {"id": "c-ibn_sina-01-starter-03", "uid": "c069", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "prompt_sha256_by_order": ["278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442"], "request_sha256": "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "compiled_prompt_sha256_by_order": ["02824b55841c1a917666db92f2dab2f28df3699581fc36e4d28cc000f8f4a105", "3ecad7421cb8e33b4a59fe7c274de9748170c6ee9f704a61867d212b5e75f8e3", "7bbe1ebe15ede610a3ec8498bcdec57585731b432ebb7110a5f114d85cea1a5d", "ab0eb080710e73e679430a3eaba59d08e0a8167fd5d813ba707d610bc197b462", "042751ebc05e9d1454186cdd6852bc9b07fe294fe08da90a89e37231cf9d98cf", "3f9029cec46662495600d1afce7d6d5af2b636cd3b5b845e97cd92ab4ec9e17e"], "compiled_tokens_by_order": [355, 355, 355, 355, 355, 355], "choice": "ibn_sina", "probs": {"darwin": 0.014328222497878532, "einstein": 0.014328222497878532, "hypatia": 0.09343174722451043, "ibn_sina": 0.7822944416474449, "sagan": 0.05666921928218631, "socrates": 0.03894814685010129}, "confidence": 0.5399530552978797, "seconds": 0.17323010266591155, "seconds_all_orders": 1.0393806159954693, "seconds_by_order": [0.17104739799833624, 0.17564147200027946, 0.17121088699786924, 0.173966604997986, 0.17252552599893534, 0.174988728002063], "argmax_stable": false, "spread": {"darwin": 0.010869118590608755, "einstein": 0.016440061945583312, "hypatia": 0.46975304877890367, "ibn_sina": 0.5513369937142609, "sagan": 0.06970765644944064, "socrates": 0.033701328295193435}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.4700000000000001, "ibn_sina": 0.55, "sagan": 0.07, "socrates": 0.03}, "choices_by_order": ["ibn_sina", "hypatia", "ibn_sina", "ibn_sina", "ibn_sina", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.014328222497878532, "einstein": 0.014328222497878532, "hypatia": 0.09343174722451043, "ibn_sina": 0.7822944416474449, "sagan": 0.05666921928218631, "socrates": 0.03894814685010129}, {"ibn_sina": 0.30145109842896356, "hypatia": 0.5631847960034141, "sagan": 0.076218773881676, "socrates": 0.03382187588767179, "darwin": 0.010315089351223957, "einstein": 0.01500836644705049}, {"hypatia": 0.11541231799101385, "darwin": 0.00574604096631226, "sagan": 0.006511117432235349, "ibn_sina": 0.8527880921432245, "einstein": 0.012164368821140165, "socrates": 0.0073780626460739325}, {"ibn_sina": 0.6645131012432169, "einstein": 0.012955942897741261, "darwin": 0.00507362013765879, "sagan": 0.017708674316313934, "hypatia": 0.2770102734870765, "socrates": 0.022738387917992504}, {"einstein": 0.02351460937157749, "sagan": 0.018313196192202747, "socrates": 0.005246818554907854, "ibn_sina": 0.7315180557759745, "darwin": 0.011823888135680586, "hypatia": 0.20958343196965679}, {"socrates": 0.015942738728267545, "ibn_sina": 0.677902500770241, "hypatia": 0.22008271950345432, "einstein": 0.00707454742599418, "sagan": 0.06305475484377539, "darwin": 0.015942738728267545}], "prompt_tokens": 355, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.014328222497878532, "einstein": 0.014328222497878532, "hypatia": 0.09343174722451043, "ibn_sina": 0.7822944416474449, "sagan": 0.05666921928218631, "socrates": 0.03894814685010129}, "confidence": 0.5399530552978797, "choice": "ibn_sina"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.01656799786724, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "02824b55841c1a917666db92f2dab2f28df3699581fc36e4d28cc000f8f4a105", "compiled_ids_sha256": "052585d008f125e9a9c56574e6532af6ea4dffeaa060a8ab415742f7363867e1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 15.125, 17.0, 19.125, 16.5, 16.125], "runtime_probabilities": [0.01432822272181511, 0.01432822272181511, 0.0934317484498024, 0.7822943925857544, 0.056669220328330994, 0.038948144763708115], "decide_seconds": 0.16678897999736364, "probabilities_f64_minus_runtime_max_abs": 4.906169048801701e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.30145109842896356, "hypatia": 0.5631847960034141, "sagan": 0.076218773881676, "socrates": 0.03382187588767179, "darwin": 0.010315089351223957, "einstein": 0.01500836644705049}, "confidence": 0.3828515014335827, "choice": "hypatia"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 175.61529200247605, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "3ecad7421cb8e33b4a59fe7c274de9748170c6ee9f704a61867d212b5e75f8e3", "compiled_ids_sha256": "f95fa995b9c8fb33c4c91228baf13be38103e65eb2ccc52cdc0e9fbefbf430c9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.0, 18.625, 16.625, 15.8125, 14.625, 15.0], "runtime_probabilities": [0.3014511168003082, 0.5631847977638245, 0.07621876895427704, 0.03382187336683273, 0.010315089486539364, 0.015008365735411644], "decide_seconds": 0.17147012500208803, "probabilities_f64_minus_runtime_max_abs": 1.8371344667755807e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.11541231799101385, "darwin": 0.00574604096631226, "sagan": 0.006511117432235349, "ibn_sina": 0.8527880921432245, "einstein": 0.012164368821140165, "socrates": 0.0073780626460739325}, "confidence": 0.7001354805170115, "choice": "ibn_sina"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.18445700180018, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "7bbe1ebe15ede610a3ec8498bcdec57585731b432ebb7110a5f114d85cea1a5d", "compiled_ids_sha256": "dc68a45dcb327e734dfecde766161ceaa2eafbddf62c52cedae022d96a06b1b2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 14.375, 14.5, 19.375, 15.125, 14.625], "runtime_probabilities": [0.11541230976581573, 0.005746040493249893, 0.006511116400361061, 0.8527880311012268, 0.012164367362856865, 0.007378062233328819], "decide_seconds": 0.16704713799845194, "probabilities_f64_minus_runtime_max_abs": 6.104199767875684e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.6645131012432169, "einstein": 0.012955942897741261, "darwin": 0.00507362013765879, "sagan": 0.017708674316313934, "hypatia": 0.2770102734870765, "socrates": 0.022738387917992504}, "confidence": 0.5156891222082298, "choice": "ibn_sina"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 173.939356001938, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "ab0eb080710e73e679430a3eaba59d08e0a8167fd5d813ba707d610bc197b462", "compiled_ids_sha256": "7bfff541c49a042871015082fdcb64cc955c8134ad0d1b79e159a2e7eded1049", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 15.1875, 14.25, 15.5, 18.25, 15.75], "runtime_probabilities": [0.664513111114502, 0.012955943122506142, 0.00507362000644207, 0.0177086740732193, 0.2770102620124817, 0.022738387808203697], "decide_seconds": 0.169622679000895, "probabilities_f64_minus_runtime_max_abs": 1.147459482986335e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.02351460937157749, "sagan": 0.018313196192202747, "socrates": 0.005246818554907854, "ibn_sina": 0.7315180557759745, "darwin": 0.011823888135680586, "hypatia": 0.20958343196965679}, "confidence": 0.5548209650262659, "choice": "ibn_sina"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 172.4985660002858, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "042751ebc05e9d1454186cdd6852bc9b07fe294fe08da90a89e37231cf9d98cf", "compiled_ids_sha256": "fe286b5f239c8228fcb9ad0e7d91d3e03e70bebc6af01eb2b5f7934b430bd552", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 15.6875, 14.4375, 19.375, 15.25, 18.125], "runtime_probabilities": [0.023514609783887863, 0.018313197419047356, 0.0052468180656433105, 0.731518030166626, 0.011823887936770916, 0.20958344638347626], "decide_seconds": 0.16824003999863635, "probabilities_f64_minus_runtime_max_abs": 2.5609348530686304e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015942738728267545, "ibn_sina": 0.677902500770241, "hypatia": 0.22008271950345432, "einstein": 0.00707454742599418, "sagan": 0.06305475484377539, "darwin": 0.015942738728267545}, "confidence": 0.47652121450833274, "choice": "ibn_sina"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 174.96285800007172, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "3f9029cec46662495600d1afce7d6d5af2b636cd3b5b845e97cd92ab4ec9e17e", "compiled_ids_sha256": "a9229864e6acd49900507f8a23c861d2faedba6f3f76c71372044447532b1a1b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 18.875, 17.75, 14.3125, 16.5, 15.125], "runtime_probabilities": [0.015942739322781563, 0.6779024600982666, 0.22008271515369415, 0.007074547465890646, 0.06305474787950516, 0.015942739322781563], "decide_seconds": 0.1708313389972318, "probabilities_f64_minus_runtime_max_abs": 4.0671974343986506e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:06Z"} {"id": "c-ibn_sina-01-starter-03", "uid": "c070", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "prompt_sha256_by_order": ["1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728"], "request_sha256": "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "compiled_prompt_sha256_by_order": ["646ca9742583b77ea42ed581986481a7143ca23eac7d883e38862c3cb4e2dc56", "dc1c333565974be045f36a600d54c50e1aa495f573dc2b17ace93b9a2c7938bf", "ecd1b862997cce847636020e0e726395ee6c9500788bcaf5b73174dd05960a9a", "02f3ff7f548a223def5e7215730f9572be0e74a03cb59652d7695138f0a8ff85", "be65ec3c9a3389e1d13c475134d8ad45cac3e1c23f304e74cdb267d6d7747916", "834fd24ef1341869a67187414c18d1d9d11e271351083da69d4bd1f327e9c426"], "compiled_tokens_by_order": [371, 371, 371, 371, 371, 371], "choice": "sagan", "probs": {"darwin": 0.01381381143751098, "einstein": 0.07949299626413925, "hypatia": 0.017737284987095925, "ibn_sina": 0.005081817232079218, "sagan": 0.8546302510364795, "socrates": 0.029243839042695098}, "confidence": 0.6671771761409727, "seconds": 0.1749267881665825, "seconds_all_orders": 1.049560728999495, "seconds_by_order": [0.17954805299814325, 0.17870740899888915, 0.1709587690020271, 0.17155250400173827, 0.17098244799854, 0.17781154600015725], "argmax_stable": true, "spread": {"darwin": 0.09512083700811268, "einstein": 0.20335707937585767, "hypatia": 0.017002645581765778, "ibn_sina": 0.012198485682253827, "sagan": 0.30636683831613243, "socrates": 0.08301520109536462}, "spread_2dp": {"darwin": 0.1, "einstein": 0.21, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.30000000000000004, "socrates": 0.09000000000000001}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01381381143751098, "einstein": 0.07949299626413925, "hypatia": 0.017737284987095925, "ibn_sina": 0.005081817232079218, "sagan": 0.8546302510364795, "socrates": 0.029243839042695098}, {"ibn_sina": 0.013157334981904375, "hypatia": 0.007042613917917019, "sagan": 0.5955464553497851, "socrates": 0.05539444934558148, "darwin": 0.0805984482153239, "einstein": 0.24826069818948807}, {"hypatia": 0.011876566927281491, "darwin": 0.012642539685058254, "sagan": 0.6902592785247907, "ibn_sina": 0.017280302914333044, "einstein": 0.174524876982684, "socrates": 0.09341643496585249}, {"ibn_sina": 0.015514704509917354, "einstein": 0.057644120320310605, "darwin": 0.057644120320310605, "sagan": 0.7957525353523055, "hypatia": 0.02257376186226804, "socrates": 0.050870757634887846}, {"einstein": 0.13837091466212684, "sagan": 0.620135415893935, "socrates": 0.09510084614378032, "ibn_sina": 0.014584187107304074, "darwin": 0.10776337669317093, "hypatia": 0.024045259499682798}, {"socrates": 0.012085645048415704, "ibn_sina": 0.005708854489171641, "hypatia": 0.011353412831008715, "einstein": 0.044903618813630385, "sagan": 0.9019132936659175, "darwin": 0.024035175151856006}], "prompt_tokens": 371, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01381381143751098, "einstein": 0.07949299626413925, "hypatia": 0.017737284987095925, "ibn_sina": 0.005081817232079218, "sagan": 0.8546302510364795, "socrates": 0.029243839042695098}, "confidence": 0.6671771761409727, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 179.51887299932423, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "646ca9742583b77ea42ed581986481a7143ca23eac7d883e38862c3cb4e2dc56", "compiled_ids_sha256": "823c9e43923cf82515b47c19a4962b220c53c5e9d09cfb9b90c5cd78de0ad4e1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 17.0, 15.5, 14.25, 19.375, 16.0], "runtime_probabilities": [0.013813811354339123, 0.07949299365282059, 0.017737284302711487, 0.005081817042082548, 0.8546302914619446, 0.029243838042020798], "decide_seconds": 0.17504832700069528, "probabilities_f64_minus_runtime_max_abs": 4.042546508653544e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013157334981904375, "hypatia": 0.007042613917917019, "sagan": 0.5955464553497851, "socrates": 0.05539444934558148, "darwin": 0.0805984482153239, "einstein": 0.24826069818948807}, "confidence": 0.3806776718383611, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 178.68210000233375, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "dc1c333565974be045f36a600d54c50e1aa495f573dc2b17ace93b9a2c7938bf", "compiled_ids_sha256": "a687eab0099bf1295de853b61258082b62c7a027819d9f9c7c075af9aef4dda9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.8125, 19.25, 16.875, 17.25, 18.375], "runtime_probabilities": [0.013157335110008717, 0.007042613811790943, 0.5955464839935303, 0.05539444833993912, 0.08059845119714737, 0.2482607066631317], "decide_seconds": 0.17429016299865907, "probabilities_f64_minus_runtime_max_abs": 2.8643745153544842e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.011876566927281491, "darwin": 0.012642539685058254, "sagan": 0.6902592785247907, "ibn_sina": 0.017280302914333044, "einstein": 0.174524876982684, "socrates": 0.09341643496585249}, "confidence": 0.464195720940907, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 170.933088997117, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "ecd1b862997cce847636020e0e726395ee6c9500788bcaf5b73174dd05960a9a", "compiled_ids_sha256": "a2e3a44700654256d50354a44333f6a7dca60e7c3386f60b83a69af8b65084c6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 15.5, 19.5, 15.8125, 18.125, 17.5], "runtime_probabilities": [0.011876567266881466, 0.012642540037631989, 0.6902592778205872, 0.017280304804444313, 0.17452487349510193, 0.09341643750667572], "decide_seconds": 0.1666146019997541, "probabilities_f64_minus_runtime_max_abs": 3.487582084238383e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.015514704509917354, "einstein": 0.057644120320310605, "darwin": 0.057644120320310605, "sagan": 0.7957525353523055, "hypatia": 0.02257376186226804, "socrates": 0.050870757634887846}, "confidence": 0.5465343247896235, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 171.52772400004324, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "02f3ff7f548a223def5e7215730f9572be0e74a03cb59652d7695138f0a8ff85", "compiled_ids_sha256": "d1e1e5f6a028d1c32b396256d6d506bcee4d2a16aac309a950f1455b92efe095", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 16.875, 16.875, 19.5, 15.9375, 16.75], "runtime_probabilities": [0.015514706261456013, 0.05764412507414818, 0.05764412507414818, 0.7957525849342346, 0.022573763504624367, 0.050870757550001144], "decide_seconds": 0.1672581570019247, "probabilities_f64_minus_runtime_max_abs": 4.958192911796999e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.13837091466212684, "sagan": 0.620135415893935, "socrates": 0.09510084614378032, "ibn_sina": 0.014584187107304074, "darwin": 0.10776337669317093, "hypatia": 0.024045259499682798}, "confidence": 0.33857663644875036, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 170.95737799900235, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "be65ec3c9a3389e1d13c475134d8ad45cac3e1c23f304e74cdb267d6d7747916", "compiled_ids_sha256": "533477d5266939ad3a59208b2c821c0985ab669161926d2be619429ebc2e84a7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.75, 19.25, 17.375, 15.5, 17.5, 16.0], "runtime_probabilities": [0.13837091624736786, 0.6201353669166565, 0.09510084241628647, 0.014584186486899853, 0.1077633649110794, 0.02404525876045227], "decide_seconds": 0.16653627299820073, "probabilities_f64_minus_runtime_max_abs": 4.897727845509792e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.012085645048415704, "ibn_sina": 0.005708854489171641, "hypatia": 0.011353412831008715, "einstein": 0.044903618813630385, "sagan": 0.9019132936659175, "darwin": 0.024035175151856006}, "confidence": 0.7456316733644556, "choice": "sagan"}}, "usage": {"input_tokens": 371, "output_tokens": 1}, "latency_ms": 177.78635600188863, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 371, "compiled_prompt_sha256": "834fd24ef1341869a67187414c18d1d9d11e271351083da69d4bd1f327e9c426", "compiled_ids_sha256": "02b9fa41fcedaaecb6d6a8c2bf217b6882610354beef3fc1ecf54b000ca77d04", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 14.4375, 15.125, 16.5, 19.5, 15.875], "runtime_probabilities": [0.012085646390914917, 0.005708855111151934, 0.011353414505720139, 0.04490362107753754, 0.9019134044647217, 0.024035178124904633], "decide_seconds": 0.1735229090008943, "probabilities_f64_minus_runtime_max_abs": 1.1079880413156218e-07}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:07Z"} {"id": "c-ibn_sina-01-starter-03", "uid": "c071", "task": "kperm", "label": "ibn_sina", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "prompt_sha256_by_order": ["ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c"], "request_sha256": "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "compiled_prompt_sha256_by_order": ["672de484d211354cd558d5df735a439daa860248a900b0c7bbcb4c54d1e2256c", "a8128f28a15ff483622d517bc12f2ed8a5ab0143b3fa4fe4aa64156368f241cd", "5100a77d8b8aaf9bc8b5eef5d31b99e6b8a3d9209c5a84cd1290982a66bc2a58", "fe997899d73ccbd8dca794dbf8f689c833e0142e0a78d35b1ef88ef4ab9def34", "75e050785f2514cb05d9f062d6aff813c2d8c1e32dbb1a08a292d29024926478", "f29fd6a1c397ef3f4513b5168cd48c0fb88a8c198617030be4be406047fe3896"], "compiled_tokens_by_order": [346, 346, 346, 346, 346, 346], "choice": "sagan", "probs": {"darwin": 0.007154552435925905, "einstein": 0.004339455408910769, "hypatia": 0.22257160240993945, "ibn_sina": 0.008107170025154934, "sagan": 0.6855687998337805, "socrates": 0.07225841988628845}, "confidence": 0.50826679661462, "seconds": 0.1716445118339228, "seconds_all_orders": 1.0298670710035367, "seconds_by_order": [0.17189592200156767, 0.17088190000140457, 0.17072453099899576, 0.17292634299883503, 0.1726681149993965, 0.17077026000333717], "argmax_stable": false, "spread": {"darwin": 0.008467059018214634, "einstein": 0.029722400151614987, "hypatia": 0.7188444363408015, "ibn_sina": 0.042982051372529265, "sagan": 0.7588948431028649, "socrates": 0.08433946534874832}, "spread_2dp": {"darwin": 0.0, "einstein": 0.03, "hypatia": 0.72, "ibn_sina": 0.04, "sagan": 0.76, "socrates": 0.09000000000000001}, "choices_by_order": ["sagan", "hypatia", "sagan", "hypatia", "hypatia", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.007154552435925905, "einstein": 0.004339455408910769, "hypatia": 0.22257160240993945, "ibn_sina": 0.008107170025154934, "sagan": 0.6855687998337805, "socrates": 0.07225841988628845}, {"ibn_sina": 0.031623267442398156, "hypatia": 0.5605358275229169, "sagan": 0.2647783765348322, "socrates": 0.09740652119391584, "darwin": 0.014032739863538774, "einstein": 0.031623267442398156}, {"hypatia": 0.28515540764441527, "darwin": 0.005918211676764777, "sagan": 0.5327404375716163, "ibn_sina": 0.04955257968258925, "einstein": 0.03405695675227895, "socrates": 0.09257640667233544}, {"ibn_sina": 0.02201269475854643, "einstein": 0.004334556600663959, "darwin": 0.00556568084532414, "sagan": 0.44213679335164396, "hypatia": 0.5010066234303424, "socrates": 0.02494365101347912}, {"einstein": 0.011052813930736793, "sagan": 0.049535275389863395, "socrates": 0.0385781112632833, "ibn_sina": 0.014192094012986383, "darwin": 0.008607940144400347, "hypatia": 0.8780337652587298}, {"socrates": 0.01306705584516751, "ibn_sina": 0.006570528310059986, "hypatia": 0.15918932891792822, "einstein": 0.0061724401240561035, "sagan": 0.8084301184927283, "darwin": 0.006570528310059986}], "prompt_tokens": 346, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.007154552435925905, "einstein": 0.004339455408910769, "hypatia": 0.22257160240993945, "ibn_sina": 0.008107170025154934, "sagan": 0.6855687998337805, "socrates": 0.07225841988628845}, "confidence": 0.50826679661462, "choice": "sagan"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 171.86916200080304, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "672de484d211354cd558d5df735a439daa860248a900b0c7bbcb4c54d1e2256c", "compiled_ids_sha256": "1d298de3fe49f6c758af9b8bfaf818d27668637424e77a80f4a96f9bfce9406e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 14.3125, 18.25, 14.9375, 19.375, 17.125], "runtime_probabilities": [0.007154553197324276, 0.0043394556269049644, 0.22257161140441895, 0.008107170462608337, 0.6855688691139221, 0.07225842028856277], "decide_seconds": 0.1676703439989069, "probabilities_f64_minus_runtime_max_abs": 6.928014162355112e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.031623267442398156, "hypatia": 0.5605358275229169, "sagan": 0.2647783765348322, "socrates": 0.09740652119391584, "darwin": 0.014032739863538774, "einstein": 0.031623267442398156}, "confidence": 0.3405996459416739, "choice": "hypatia"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 170.85686000064015, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "a8128f28a15ff483622d517bc12f2ed8a5ab0143b3fa4fe4aa64156368f241cd", "compiled_ids_sha256": "1203a8a4a430230d9427ef458f1b56a6a1d3470965f9dde2d9da5b286cc41170", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 19.125, 18.375, 17.375, 15.4375, 16.25], "runtime_probabilities": [0.031623270362615585, 0.5605358481407166, 0.26477837562561035, 0.09740651398897171, 0.014032740145921707, 0.031623270362615585], "decide_seconds": 0.1667367909976747, "probabilities_f64_minus_runtime_max_abs": 2.061779968975941e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.28515540764441527, "darwin": 0.005918211676764777, "sagan": 0.5327404375716163, "ibn_sina": 0.04955257968258925, "einstein": 0.03405695675227895, "socrates": 0.09257640667233544}, "confidence": 0.32584268150414364, "choice": "sagan"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 170.69963099856977, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "5100a77d8b8aaf9bc8b5eef5d31b99e6b8a3d9209c5a84cd1290982a66bc2a58", "compiled_ids_sha256": "a1aaf0ac4619ace6345bedd8e90dfa5e04ac6d0a393c5629eb6b287fab5bde90", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.875, 15.0, 19.5, 17.125, 16.75, 17.75], "runtime_probabilities": [0.28515544533729553, 0.0059182122349739075, 0.5327404737472534, 0.04955257847905159, 0.03405695781111717, 0.09257639944553375], "decide_seconds": 0.1666277819967945, "probabilities_f64_minus_runtime_max_abs": 3.76928802614529e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02201269475854643, "einstein": 0.004334556600663959, "darwin": 0.00556568084532414, "sagan": 0.44213679335164396, "hypatia": 0.5010066234303424, "socrates": 0.02494365101347912}, "confidence": 0.4777991105274363, "choice": "hypatia"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 172.902394002449, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "fe997899d73ccbd8dca794dbf8f689c833e0142e0a78d35b1ef88ef4ab9def34", "compiled_ids_sha256": "846b5ae6d8fd6af7ec5f0eb90889786616d7df0f00cb702678789880ff8121b5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 14.5, 14.75, 19.125, 19.25, 16.25], "runtime_probabilities": [0.02201269380748272, 0.0043345564045012, 0.005565681029111147, 0.4421367645263672, 0.5010066032409668, 0.024943649768829346], "decide_seconds": 0.16887082500034012, "probabilities_f64_minus_runtime_max_abs": 2.882527677661173e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.011052813930736793, "sagan": 0.049535275389863395, "socrates": 0.0385781112632833, "ibn_sina": 0.014192094012986383, "darwin": 0.008607940144400347, "hypatia": 0.8780337652587298}, "confidence": 0.6987592784055292, "choice": "hypatia"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 172.64284499833593, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "75e050785f2514cb05d9f062d6aff813c2d8c1e32dbb1a08a292d29024926478", "compiled_ids_sha256": "7a38008ababcb3728afc87183afc13ede283248e45f4595fecc5945af733d925", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 16.75, 16.5, 15.5, 15.0, 19.625], "runtime_probabilities": [0.011052814312279224, 0.04953527823090553, 0.03857811167836189, 0.014192093163728714, 0.008607940748333931, 0.8780337572097778], "decide_seconds": 0.16836145900015254, "probabilities_f64_minus_runtime_max_abs": 8.048951949390926e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.01306705584516751, "ibn_sina": 0.006570528310059986, "hypatia": 0.15918932891792822, "einstein": 0.0061724401240561035, "sagan": 0.8084301184927283, "darwin": 0.006570528310059986}, "confidence": 0.6547654451984135, "choice": "sagan"}}, "usage": {"input_tokens": 346, "output_tokens": 1}, "latency_ms": 170.74546000003465, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 346, "compiled_prompt_sha256": "f29fd6a1c397ef3f4513b5168cd48c0fb88a8c198617030be4be406047fe3896", "compiled_ids_sha256": "dcb3b20454214f3da3cbe2f735f49a5739f43117058d02cc952bbd3445a4b6d8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 14.6875, 17.875, 14.625, 19.5, 14.6875], "runtime_probabilities": [0.013067055493593216, 0.006570528727024794, 0.15918931365013123, 0.00617243954911828, 0.8084301352500916, 0.006570528727024794], "decide_seconds": 0.16666397100198083, "probabilities_f64_minus_runtime_max_abs": 1.6757363274777504e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:08Z"} {"id": "c-sagan-01-dessert-01", "uid": "c072", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "prompt_sha256_by_order": ["66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f"], "request_sha256": "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "compiled_prompt_sha256_by_order": ["63461712757b659f4c1e81c96db955b9576be9a529dd979a61ae859b5e4fe20e", "fef35ee41e526b7e88efc66b7cbdab640adbb8dbcd04a746d68aff571097b885", "99da7fcb270f502a8f50833875aab41f4abac2a90246adecdf432875f9252bfa", "d0dc067a75fd89123503cfa14e73711e76c13003de8aa84a7f7dbd72215d1077", "ca62948d8bf67ae01de708aec872f941b3562a277a11dd49f10b1f4a2ee701dc", "26507bfc7594cde5fde7bf393f257cfff45aab5996b11561a7846383af313423"], "compiled_tokens_by_order": [404, 404, 404, 404, 404, 404], "choice": "sagan", "probs": {"darwin": 0.016614256520079966, "einstein": 0.21545695682713795, "hypatia": 0.042425990303229445, "ibn_sina": 0.01377370243206486, "sagan": 0.6636541486355552, "socrates": 0.04807494528193262}, "confidence": 0.43636989673848803, "seconds": 0.18015247516693003, "seconds_all_orders": 1.0809148510015802, "seconds_by_order": [0.17763593800191302, 0.1833036739990348, 0.17738438000014867, 0.1777097070007585, 0.18061093500000425, 0.18427021699972101], "argmax_stable": false, "spread": {"darwin": 0.1012073933914589, "einstein": 0.6023767629739949, "hypatia": 0.09327793296826628, "ibn_sina": 0.056390158866651725, "sagan": 0.6322889302787625, "socrates": 0.23396868422949696}, "spread_2dp": {"darwin": 0.1, "einstein": 0.61, "hypatia": 0.09, "ibn_sina": 0.060000000000000005, "sagan": 0.64, "socrates": 0.23}, "choices_by_order": ["sagan", "einstein", "sagan", "sagan", "einstein", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.016614256520079966, "einstein": 0.21545695682713795, "hypatia": 0.042425990303229445, "ibn_sina": 0.01377370243206486, "sagan": 0.6636541486355552, "socrates": 0.04807494528193262}, {"ibn_sina": 0.025399819597287297, "hypatia": 0.021057190042680335, "sagan": 0.3973200278851538, "socrates": 0.069043868057444, "darwin": 0.03695651944690415, "einstein": 0.4502225749705303}, {"hypatia": 0.017376561065870055, "darwin": 0.008737483620459856, "sagan": 0.652050643462675, "ibn_sina": 0.06872563219603063, "einstein": 0.16486422113732763, "socrates": 0.0882454585176367}, {"ibn_sina": 0.05840813550107511, "einstein": 0.12365002382609136, "darwin": 0.06618508838956084, "sagan": 0.3808688867582426, "hypatia": 0.10912076303103704, "socrates": 0.26176710249399304}, {"einstein": 0.6254224611546201, "sagan": 0.08464172592289991, "socrates": 0.20304494515517904, "ibn_sina": 0.01888612186041616, "darwin": 0.051337801863232384, "hypatia": 0.016666944043652477}, {"socrates": 0.027798418264496097, "ibn_sina": 0.012335473329378906, "hypatia": 0.10994487701191875, "einstein": 0.023045698180625133, "sagan": 0.7169306562016624, "darwin": 0.10994487701191875}], "prompt_tokens": 404, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.016614256520079966, "einstein": 0.21545695682713795, "hypatia": 0.042425990303229445, "ibn_sina": 0.01377370243206486, "sagan": 0.6636541486355552, "socrates": 0.04807494528193262}, "confidence": 0.43636989673848803, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.6070880005136, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "63461712757b659f4c1e81c96db955b9576be9a529dd979a61ae859b5e4fe20e", "compiled_ids_sha256": "82d8ecc5cf442b09faecb1a0ed1d65f9bc9153c91f2b0ea1a85d3471da004e30", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 18.0, 16.375, 15.25, 19.125, 16.5], "runtime_probabilities": [0.01661425642669201, 0.21545694768428802, 0.042425986379384995, 0.013773701153695583, 0.6636541485786438, 0.04807494580745697], "decide_seconds": 0.17289909300234285, "probabilities_f64_minus_runtime_max_abs": 9.142849921595086e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.025399819597287297, "hypatia": 0.021057190042680335, "sagan": 0.3973200278851538, "socrates": 0.069043868057444, "darwin": 0.03695651944690415, "einstein": 0.4502225749705303}, "confidence": 0.326338769940798, "choice": "einstein"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 183.27813499854528, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "fef35ee41e526b7e88efc66b7cbdab640adbb8dbcd04a746d68aff571097b885", "compiled_ids_sha256": "4f9580e184738f4c2a44c36a6721b01b4fa56d35156e11683b66dc296b6ce274", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 15.4375, 18.375, 16.625, 16.0, 18.5], "runtime_probabilities": [0.025399820879101753, 0.021057190373539925, 0.3973200023174286, 0.06904386729001999, 0.03695651888847351, 0.45022255182266235], "decide_seconds": 0.17863514900091104, "probabilities_f64_minus_runtime_max_abs": 2.5567725214781234e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.017376561065870055, "darwin": 0.008737483620459856, "sagan": 0.652050643462675, "ibn_sina": 0.06872563219603063, "einstein": 0.16486422113732763, "socrates": 0.0882454585176367}, "confidence": 0.3938274694680092, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.35801999879186, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "99da7fcb270f502a8f50833875aab41f4abac2a90246adecdf432875f9252bfa", "compiled_ids_sha256": "315975be997bff408243c7754221b54a4cff448851180b92ed3c48f639fff0f2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 14.9375, 19.25, 17.0, 17.875, 17.25], "runtime_probabilities": [0.01737656071782112, 0.008737483993172646, 0.6520506143569946, 0.06872563064098358, 0.1648642122745514, 0.0882454589009285], "decide_seconds": 0.17270191500210785, "probabilities_f64_minus_runtime_max_abs": 2.9105680421359637e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.05840813550107511, "einstein": 0.12365002382609136, "darwin": 0.06618508838956084, "sagan": 0.3808688867582426, "hypatia": 0.10912076303103704, "socrates": 0.26176710249399304}, "confidence": 0.12694244290861967, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.6836170029128, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "d0dc067a75fd89123503cfa14e73711e76c13003de8aa84a7f7dbd72215d1077", "compiled_ids_sha256": "66a30b322521e26134a99c15a39e62d7cd5ccabbef5ad4b602caf86fa2599a76", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 17.125, 16.5, 18.25, 17.0, 17.875], "runtime_probabilities": [0.058408137410879135, 0.12365001440048218, 0.06618507951498032, 0.3808688819408417, 0.10912076383829117, 0.26176708936691284], "decide_seconds": 0.17307667299974128, "probabilities_f64_minus_runtime_max_abs": 1.3127080200625585e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.6254224611546201, "sagan": 0.08464172592289991, "socrates": 0.20304494515517904, "ibn_sina": 0.01888612186041616, "darwin": 0.051337801863232384, "hypatia": 0.016666944043652477}, "confidence": 0.37385538768887205, "choice": "einstein"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 180.58373499661684, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "ca62948d8bf67ae01de708aec872f941b3562a277a11dd49f10b1f4a2ee701dc", "compiled_ids_sha256": "d6e1ca776cdb5d092a1cc96e92ced5e56d48b4b1f8b94519859f603a79983a6e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.0, 17.0, 17.875, 15.5, 16.5, 15.375], "runtime_probabilities": [0.625422477722168, 0.08464173227548599, 0.20304495096206665, 0.018886122852563858, 0.05133780464529991, 0.016666945070028305], "decide_seconds": 0.1757962910014612, "probabilities_f64_minus_runtime_max_abs": 1.6567547889145828e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.027798418264496097, "ibn_sina": 0.012335473329378906, "hypatia": 0.10994487701191875, "einstein": 0.023045698180625133, "sagan": 0.7169306562016624, "darwin": 0.10994487701191875}, "confidence": 0.4615645852669823, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 184.24483700073324, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "26507bfc7594cde5fde7bf393f257cfff45aab5996b11561a7846383af313423", "compiled_ids_sha256": "525bba058f24dbe6ae5fb76c69670ab1fda2cbeb296202081b3971070bf23601", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 14.9375, 17.125, 15.5625, 19.0, 17.125], "runtime_probabilities": [0.02779841609299183, 0.01233547180891037, 0.10994487255811691, 0.023045698180794716, 0.716930627822876, 0.10994487255811691], "decide_seconds": 0.17962451199855423, "probabilities_f64_minus_runtime_max_abs": 2.837878643191516e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:09Z"} {"id": "c-sagan-01-dessert-01", "uid": "c073", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "prompt_sha256_by_order": ["985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549"], "request_sha256": "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "compiled_prompt_sha256_by_order": ["929aa8fbc3eee78c7018d5c8c61cbd18f10472b9ec10356a479f5560f36dc969", "c26c7a1876342961a9b16ea85b6008c8b4e2186347b4446ab8e8521fd9f47803", "937544d454a03b10ebdf949ac5cb53a99a436501a7429042faf1e639a219d52b", "352878a3b71191b84b3b4ab58ab21aea718d624cc86267f5f3136bf77b25d176", "aa75f9e6afad585ff0e378f8cf32e706fb131e8b9a464c07e76d9ff215412a7a", "855f48f09ae5a5443c7bc7e73391c1f4b3f2c7afa08c7a3c7b95df727bc83f3d"], "compiled_tokens_by_order": [309, 309, 309, 309, 309, 309], "choice": "sagan", "probs": {"darwin": 0.025213011525174245, "einstein": 0.01349555256623053, "hypatia": 0.022250404575795975, "ibn_sina": 0.0043813644389910505, "sagan": 0.8349402718958633, "socrates": 0.0997193949979449}, "confidence": 0.642879869045457, "seconds": 0.16812356516614577, "seconds_all_orders": 1.0087413909968745, "seconds_by_order": [0.17178637200049707, 0.16546840099908877, 0.16933105099815293, 0.16796771100052865, 0.16874959499909892, 0.1654382609995082], "argmax_stable": false, "spread": {"darwin": 0.15830661242916147, "einstein": 0.13155929328098462, "hypatia": 0.021307340747951313, "ibn_sina": 0.013473953231766988, "sagan": 0.855410528184817, "socrates": 0.5502055483323934}, "spread_2dp": {"darwin": 0.16, "einstein": 0.13, "hypatia": 0.02, "ibn_sina": 0.02, "sagan": 0.85, "socrates": 0.5499999999999999}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.025213011525174245, "einstein": 0.01349555256623053, "hypatia": 0.022250404575795975, "ibn_sina": 0.0043813644389910505, "sagan": 0.8349402718958633, "socrates": 0.0997193949979449}, {"ibn_sina": 0.005869894276715712, "hypatia": 0.0035602708480997917, "sagan": 0.8711695532055127, "socrates": 0.018080547252507257, "darwin": 0.029809782841107472, "einstein": 0.07150995157605705}, {"hypatia": 0.004644691795341118, "darwin": 0.003850584743311387, "sagan": 0.9422070211879904, "ibn_sina": 0.003617289607336992, "einstein": 0.013439861340926913, "socrates": 0.032240551325093084}, {"ibn_sina": 0.009642450611984679, "einstein": 0.012381151664944594, "darwin": 0.040596233411057066, "sagan": 0.8153971451201315, "hypatia": 0.011631015606723766, "socrates": 0.11035200358515838}, {"einstein": 0.1431032242365068, "sagan": 0.0867964930031734, "socrates": 0.5659842311526918, "ibn_sina": 0.01709124283910398, "darwin": 0.16215719717247284, "hypatia": 0.024867611596051106}, {"socrates": 0.015778682820298394, "ibn_sina": 0.003989475786615543, "hypatia": 0.02026022978311735, "einstein": 0.01154393095552218, "sagan": 0.9170480229090648, "darwin": 0.0313796577453818}], "prompt_tokens": 309, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.025213011525174245, "einstein": 0.01349555256623053, "hypatia": 0.022250404575795975, "ibn_sina": 0.0043813644389910505, "sagan": 0.8349402718958633, "socrates": 0.0997193949979449}, "confidence": 0.642879869045457, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 171.76001199914026, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "929aa8fbc3eee78c7018d5c8c61cbd18f10472b9ec10356a479f5560f36dc969", "compiled_ids_sha256": "9bfb6bfa3765f21c286543e6595240c7937c405557de6095941783fa77c32cf4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 15.125, 15.625, 14.0, 19.25, 17.125], "runtime_probabilities": [0.025213010609149933, 0.013495552353560925, 0.022250404581427574, 0.004381364211440086, 0.8349402546882629, 0.09971939772367477], "decide_seconds": 0.1678714920017228, "probabilities_f64_minus_runtime_max_abs": 1.7207600344626428e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.005869894276715712, "hypatia": 0.0035602708480997917, "sagan": 0.8711695532055127, "socrates": 0.018080547252507257, "darwin": 0.029809782841107472, "einstein": 0.07150995157605705}, "confidence": 0.7006882659329141, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.44468099891674, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "c26c7a1876342961a9b16ea85b6008c8b4e2186347b4446ab8e8521fd9f47803", "compiled_ids_sha256": "ba40f96a6ae1e1245d8ff0ffb55fbf494afc96c33709a38394fb5b4f98ec8236", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.25, 19.75, 15.875, 16.375, 17.25], "runtime_probabilities": [0.005869894754141569, 0.0035602711141109467, 0.8711696267127991, 0.01808054745197296, 0.02980978414416313, 0.0715099573135376], "decide_seconds": 0.16173612900092849, "probabilities_f64_minus_runtime_max_abs": 7.350728636890125e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.004644691795341118, "darwin": 0.003850584743311387, "sagan": 0.9422070211879904, "ibn_sina": 0.003617289607336992, "einstein": 0.013439861340926913, "socrates": 0.032240551325093084}, "confidence": 0.8373465568600373, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 169.3083710015344, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "937544d454a03b10ebdf949ac5cb53a99a436501a7429042faf1e639a219d52b", "compiled_ids_sha256": "105d7187ab5cfda37fbe5608beea109a687bf1cc2508d8b7b48f04a8116bcba6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.5, 20.0, 14.4375, 15.75, 16.625], "runtime_probabilities": [0.0046446919441223145, 0.0038505850825458765, 0.9422070384025574, 0.0036172899417579174, 0.013439861126244068, 0.032240550965070724], "decide_seconds": 0.1656402890002937, "probabilities_f64_minus_runtime_max_abs": 1.721456699410595e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.009642450611984679, "einstein": 0.012381151664944594, "darwin": 0.040596233411057066, "sagan": 0.8153971451201315, "hypatia": 0.011631015606723766, "socrates": 0.11035200358515838}, "confidence": 0.6145470569093372, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 167.94364100132952, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "352878a3b71191b84b3b4ab58ab21aea718d624cc86267f5f3136bf77b25d176", "compiled_ids_sha256": "cee15b0c067a810ecb8495956276aa604b36bf64a34a6106272a28249f816cd4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 15.1875, 16.375, 19.375, 15.125, 17.375], "runtime_probabilities": [0.009642450138926506, 0.012381152249872684, 0.040596235543489456, 0.8153972029685974, 0.011631015688180923, 0.11035200953483582], "decide_seconds": 0.16406466199987335, "probabilities_f64_minus_runtime_max_abs": 5.784846590639603e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.1431032242365068, "sagan": 0.0867964930031734, "socrates": 0.5659842311526918, "ibn_sina": 0.01709124283910398, "darwin": 0.16215719717247284, "hypatia": 0.024867611596051106}, "confidence": 0.2917986072012978, "choice": "socrates"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 168.72622499795398, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "aa75f9e6afad585ff0e378f8cf32e706fb131e8b9a464c07e76d9ff215412a7a", "compiled_ids_sha256": "b311a78c0a73ff1143291dedba188f07cbc6db0a8fb5cf901e2cdd35b8695325", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 16.75, 18.625, 15.125, 17.375, 15.5], "runtime_probabilities": [0.1431032121181488, 0.08679649233818054, 0.5659841895103455, 0.017091242596507072, 0.16215719282627106, 0.02486761100590229], "decide_seconds": 0.1648331260003033, "probabilities_f64_minus_runtime_max_abs": 4.1642346348957915e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015778682820298394, "ibn_sina": 0.003989475786615543, "hypatia": 0.02026022978311735, "einstein": 0.01154393095552218, "sagan": 0.9170480229090648, "darwin": 0.0313796577453818}, "confidence": 0.7733834881845757, "choice": "sagan"}}, "usage": {"input_tokens": 309, "output_tokens": 1}, "latency_ms": 165.41514100026689, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 309, "compiled_prompt_sha256": "855f48f09ae5a5443c7bc7e73391c1f4b3f2c7afa08c7a3c7b95df727bc83f3d", "compiled_ids_sha256": "b5b74e3ad737b5e5abade785558e11637266442154e1541c89bb8c4d990b96d1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 13.9375, 15.5625, 15.0, 19.375, 16.0], "runtime_probabilities": [0.015778683125972748, 0.003989475779235363, 0.02026022970676422, 0.011543931439518929, 0.9170480370521545, 0.03137965872883797], "decide_seconds": 0.16172865000044112, "probabilities_f64_minus_runtime_max_abs": 1.414308969938105e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:10Z"} {"id": "c-sagan-01-dessert-01", "uid": "c074", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "prompt_sha256_by_order": ["5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d"], "request_sha256": "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "compiled_prompt_sha256_by_order": ["c9162970d7a87671df926d763cd3c86cd736e6a98d9885f5541c96e5393b7873", "3cc5db792d343ae5d2f8338f11575cf3f6c37baa6dd8f862c962baf5496a6a64", "0d8b8e4fe95640102c23b96ccfdf039f3672b8aa493fe9554fe94967a2ca0c2b", "1c62b88c1ff6492b1059799160ba5c6d7e5e163c8c816ceab82eb21115459a88", "744cb9c68e7eec0166f194627adc5e0c0e5c6f6cd7ffd276ee8f6443bce4d387", "1aa36fbb6f910ed5794a5ea13797323b27841063359461e1f7e5ac5680285799"], "compiled_tokens_by_order": [288, 288, 288, 288, 288, 288], "choice": "sagan", "probs": {"darwin": 0.022172748359918372, "einstein": 0.01523909222910201, "hypatia": 0.07739049608561467, "ibn_sina": 0.006352598764690526, "sagan": 0.7342605831050506, "socrates": 0.14458448145562386}, "confidence": 0.5061833605175576, "seconds": 0.16608672933277072, "seconds_all_orders": 0.9965203759966244, "seconds_by_order": [0.17182441200202447, 0.1649465449991112, 0.16479318599886028, 0.16487659600170446, 0.1651864419982303, 0.16489319499669364], "argmax_stable": false, "spread": {"darwin": 0.05566644991605087, "einstein": 0.13686345054351665, "hypatia": 0.08257189061443712, "ibn_sina": 0.013889261183525927, "sagan": 0.7758666265320768, "socrates": 0.6664322884261806}, "spread_2dp": {"darwin": 0.049999999999999996, "einstein": 0.13999999999999999, "hypatia": 0.08, "ibn_sina": 0.01, "sagan": 0.78, "socrates": 0.6699999999999999}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.022172748359918372, "einstein": 0.01523909222910201, "hypatia": 0.07739049608561467, "ibn_sina": 0.006352598764690526, "sagan": 0.7342605831050506, "socrates": 0.14458448145562386}, {"ibn_sina": 0.0119942969439476, "hypatia": 0.015400982131328823, "sagan": 0.7420608754204207, "socrates": 0.06902241329044823, "darwin": 0.015400982131328823, "einstein": 0.14612045008252578}, {"hypatia": 0.011659317655773086, "darwin": 0.005173790847130148, "sagan": 0.8700978348022871, "ibn_sina": 0.005862673094916833, "einstein": 0.026274677932453805, "socrates": 0.080931705667439}, {"ibn_sina": 0.011886222690354439, "einstein": 0.009256999539009136, "darwin": 0.019597068177865863, "sagan": 0.6908204659532073, "hypatia": 0.044162674866209686, "socrates": 0.22427656877335359}, {"einstein": 0.034665724239354725, "sagan": 0.09423120827021021, "socrates": 0.6962796841786011, "ibn_sina": 0.01975193427844276, "darwin": 0.060840240763181015, "hypatia": 0.09423120827021021}, {"socrates": 0.02984739575242051, "ibn_sina": 0.009102941395387918, "hypatia": 0.06726214463579938, "einstein": 0.015008213104512764, "sagan": 0.8194206708095895, "darwin": 0.059358634302290006}], "prompt_tokens": 288, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.022172748359918372, "einstein": 0.01523909222910201, "hypatia": 0.07739049608561467, "ibn_sina": 0.006352598764690526, "sagan": 0.7342605831050506, "socrates": 0.14458448145562386}, "confidence": 0.5061833605175576, "choice": "sagan"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 171.79893299908144, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "c9162970d7a87671df926d763cd3c86cd736e6a98d9885f5541c96e5393b7873", "compiled_ids_sha256": "2e18cf1277f24da4cbb35e3c4d9ff0025b081d2a66b6e030656f384eefb90cf3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 15.0, 16.625, 14.125, 18.875, 17.25], "runtime_probabilities": [0.022172749042510986, 0.015239092521369457, 0.07739049941301346, 0.006352598778903484, 0.734260618686676, 0.14458447694778442], "decide_seconds": 0.16813467000247329, "probabilities_f64_minus_runtime_max_abs": 3.558162542383059e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0119942969439476, "hypatia": 0.015400982131328823, "sagan": 0.7420608754204207, "socrates": 0.06902241329044823, "darwin": 0.015400982131328823, "einstein": 0.14612045008252578}, "confidence": 0.5152630355182903, "choice": "sagan"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 164.92371499771252, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "3cc5db792d343ae5d2f8338f11575cf3f6c37baa6dd8f862c962baf5496a6a64", "compiled_ids_sha256": "5fe74006f150d870c016d10046b38516d94b44f599a303416c2c59649b0f4470", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 15.125, 19.0, 16.625, 15.125, 17.375], "runtime_probabilities": [0.011994296684861183, 0.015400981530547142, 0.7420608997344971, 0.06902240961790085, 0.015400981530547142, 0.14612044394016266], "decide_seconds": 0.16137409300063155, "probabilities_f64_minus_runtime_max_abs": 2.431407641889649e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.011659317655773086, "darwin": 0.005173790847130148, "sagan": 0.8700978348022871, "ibn_sina": 0.005862673094916833, "einstein": 0.026274677932453805, "socrates": 0.080931705667439}, "confidence": 0.704517456928117, "choice": "sagan"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 164.77050600224175, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "0d8b8e4fe95640102c23b96ccfdf039f3672b8aa493fe9554fe94967a2ca0c2b", "compiled_ids_sha256": "b36de7ee95b3245e3e669275c594732d3d1489a2a98f64d7a1e1f83d650b3049", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 14.375, 19.5, 14.5, 16.0, 17.125], "runtime_probabilities": [0.011659317649900913, 0.005173790734261274, 0.870097815990448, 0.005862673278898001, 0.026274677366018295, 0.08093170076608658], "decide_seconds": 0.1612499629991362, "probabilities_f64_minus_runtime_max_abs": 1.8811839086474436e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.011886222690354439, "einstein": 0.009256999539009136, "darwin": 0.019597068177865863, "sagan": 0.6908204659532073, "hypatia": 0.044162674866209686, "socrates": 0.22427656877335359}, "confidence": 0.4967756758043097, "choice": "sagan"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 164.8547859986138, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "1c62b88c1ff6492b1059799160ba5c6d7e5e163c8c816ceab82eb21115459a88", "compiled_ids_sha256": "2fc342ca47a1573e492cd12a6123939e4eca316384c14f971ac9a1ccc3ed674d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.4375, 15.1875, 18.75, 16.0, 17.625], "runtime_probabilities": [0.011886222288012505, 0.0092569999396801, 0.019597068428993225, 0.6908204555511475, 0.04416267201304436, 0.2242765575647354], "decide_seconds": 0.16129302300032577, "probabilities_f64_minus_runtime_max_abs": 1.1208618172497964e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.034665724239354725, "sagan": 0.09423120827021021, "socrates": 0.6962796841786011, "ibn_sina": 0.01975193427844276, "darwin": 0.060840240763181015, "hypatia": 0.09423120827021021}, "confidence": 0.40751522118919803, "choice": "socrates"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 165.16335400228854, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "744cb9c68e7eec0166f194627adc5e0c0e5c6f6cd7ffd276ee8f6443bce4d387", "compiled_ids_sha256": "9bcdfb63ad98fa9d39ea67de012cd2af2e9c05cec026eef1080ae7ac37f105c7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 16.375, 18.375, 14.8125, 15.9375, 16.375], "runtime_probabilities": [0.034665726125240326, 0.09423121064901352, 0.6962797045707703, 0.019751936197280884, 0.060840241611003876, 0.09423121064901352], "decide_seconds": 0.16166576000250643, "probabilities_f64_minus_runtime_max_abs": 2.039216917548714e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.02984739575242051, "ibn_sina": 0.009102941395387918, "hypatia": 0.06726214463579938, "einstein": 0.015008213104512764, "sagan": 0.8194206708095895, "darwin": 0.059358634302290006}, "confidence": 0.5964887789898166, "choice": "sagan"}}, "usage": {"input_tokens": 288, "output_tokens": 1}, "latency_ms": 164.87021500142873, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 288, "compiled_prompt_sha256": "1aa36fbb6f910ed5794a5ea13797323b27841063359461e1f7e5ac5680285799", "compiled_ids_sha256": "6b1b395d8c3d0dd87a4da79997213d77240cb646d1f13e095909fe1ed8d00e6a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.25, 16.25, 14.75, 18.75, 16.125], "runtime_probabilities": [0.0298473984003067, 0.009102942422032356, 0.06726215034723282, 0.015008213929831982, 0.8194206953048706, 0.059358637779951096], "decide_seconds": 0.16119100399737363, "probabilities_f64_minus_runtime_max_abs": 2.449528113679378e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:11Z"} {"id": "c-sagan-01-dessert-03", "uid": "c075", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "prompt_sha256_by_order": ["d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d"], "request_sha256": "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "compiled_prompt_sha256_by_order": ["efc2df06e0745877c07b202c66d4cf238051ae0a1ae157bb1d319d831d699748", "f20d1844845c6c8e4f2a196816988ad19fff335abd61838ca0450870e8b6ae25", "64b37548a0164dae355cd9c73b2d5501f8160f509954d4eb2eb51fe8ec653271", "ad23a3a76e6a573aa4f2b8864ce42d93dc025e5a1d43171c72356e91eddc1a17", "26ba57f0d7982bb838d8721008767af0625af79cf648c0800a20556c8564704e", "f83db599e2edc116912d3740c5169e74befbc33c37d53e395ea2b3a2388f1c99"], "compiled_tokens_by_order": [327, 327, 327, 327, 327, 327], "choice": "sagan", "probs": {"darwin": 0.002417585556467364, "einstein": 0.01678136992733774, "hypatia": 0.0022711114522146274, "ibn_sina": 0.0010727970875732632, "sagan": 0.9753236242110912, "socrates": 0.002133511765315704}, "confidence": 0.9208540126086644, "seconds": 0.174898348166001, "seconds_all_orders": 1.049390088996006, "seconds_by_order": [0.17804702399735106, 0.17087311900104396, 0.1758150809982908, 0.17469709999932093, 0.1782913520000875, 0.1716664129999117], "argmax_stable": true, "spread": {"darwin": 0.0030254982245152212, "einstein": 0.018830490191104648, "hypatia": 0.0016077328890294518, "ibn_sina": 0.0016635150430772848, "sagan": 0.020796867382422257, "socrates": 0.0030198724515801577}, "spread_2dp": {"darwin": 0.0, "einstein": 0.019999999999999997, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.020000000000000018, "socrates": 0.0}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.002417585556467364, "einstein": 0.01678136992733774, "hypatia": 0.0022711114522146274, "ibn_sina": 0.0010727970875732632, "sagan": 0.9753236242110912, "socrates": 0.002133511765315704}, {"ibn_sina": 0.0016420731072750157, "hypatia": 0.0011285792416210962, "sagan": 0.963872990511616, "socrates": 0.0023892022732605804, "darwin": 0.0018607126013313207, "einstein": 0.029106442264896032}, {"hypatia": 0.0014803912565015537, "darwin": 0.0009558124332969637, "sagan": 0.9846698578940383, "ibn_sina": 0.0013906988844324146, "einstein": 0.010275952073791384, "socrates": 0.001227287457939458}, {"ibn_sina": 0.002736312130650548, "einstein": 0.014792363621541068, "darwin": 0.003981310657812185, "sagan": 0.974194597804198, "hypatia": 0.002736312130650548, "socrates": 0.0015591036551475597}, {"einstein": 0.014824044679406062, "sagan": 0.9762810470164021, "socrates": 0.004247159909519615, "ibn_sina": 0.0012953105883639045, "darwin": 0.002135606119198667, "hypatia": 0.0012168316871096607}, {"socrates": 0.003107011713646269, "ibn_sina": 0.0014676484125010407, "hypatia": 0.0022731383290238985, "einstein": 0.01482272387905428, "sagan": 0.9761940618259075, "darwin": 0.002135415839867047}], "prompt_tokens": 327, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.002417585556467364, "einstein": 0.01678136992733774, "hypatia": 0.0022711114522146274, "ibn_sina": 0.0010727970875732632, "sagan": 0.9753236242110912, "socrates": 0.002133511765315704}, "confidence": 0.9208540126086644, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 178.01756400149316, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "efc2df06e0745877c07b202c66d4cf238051ae0a1ae157bb1d319d831d699748", "compiled_ids_sha256": "6a8b4cc8d5bca66563e796c67461a643efc58f4871ab7302ea9ea36cd395825a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 15.3125, 13.3125, 12.5625, 19.375, 13.25], "runtime_probabilities": [0.00241758581250906, 0.01678137108683586, 0.002271111588925123, 0.0010727971093729138, 0.9753236770629883, 0.0021335119381546974], "decide_seconds": 0.17395804499756196, "probabilities_f64_minus_runtime_max_abs": 5.285189708104099e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0016420731072750157, "hypatia": 0.0011285792416210962, "sagan": 0.963872990511616, "socrates": 0.0023892022732605804, "darwin": 0.0018607126013313207, "einstein": 0.029106442264896032}, "confidence": 0.8980225359285539, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 170.84749899731833, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "f20d1844845c6c8e4f2a196816988ad19fff335abd61838ca0450870e8b6ae25", "compiled_ids_sha256": "8a9d223ee987f87922beb172dc0b7755174f1097c7ab839b8802b411db25d650", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 12.875, 19.625, 13.625, 13.375, 16.125], "runtime_probabilities": [0.0016420732717961073, 0.001128579257056117, 0.963873028755188, 0.0023892023600637913, 0.00186071265488863, 0.029106443747878075], "decide_seconds": 0.1668676500012225, "probabilities_f64_minus_runtime_max_abs": 3.8243571975904445e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0014803912565015537, "darwin": 0.0009558124332969637, "sagan": 0.9846698578940383, "ibn_sina": 0.0013906988844324146, "einstein": 0.010275952073791384, "socrates": 0.001227287457939458}, "confidence": 0.9464658427350261, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 175.78378100006375, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "64b37548a0164dae355cd9c73b2d5501f8160f509954d4eb2eb51fe8ec653271", "compiled_ids_sha256": "e23eaa8b9eaf5627189988d8e7cc1eb79d9b86d8acf644502e198e33c739a3e1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.25, 12.8125, 19.75, 13.1875, 15.1875, 13.0625], "runtime_probabilities": [0.001480391132645309, 0.0009558123419992626, 0.9846698045730591, 0.0013906988315284252, 0.010275952517986298, 0.0012272874591872096], "decide_seconds": 0.1718911019997904, "probabilities_f64_minus_runtime_max_abs": 5.33209791875322e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.002736312130650548, "einstein": 0.014792363621541068, "darwin": 0.003981310657812185, "sagan": 0.974194597804198, "hypatia": 0.002736312130650548, "socrates": 0.0015591036551475597}, "confidence": 0.9150707256747135, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 174.67115000181366, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "ad23a3a76e6a573aa4f2b8864ce42d93dc025e5a1d43171c72356e91eddc1a17", "compiled_ids_sha256": "f479e57cbfa5839bfebcc69460d11a8ac7bede7572e2a3784529cb17fdacb7fa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 15.4375, 14.125, 19.625, 13.75, 13.1875], "runtime_probabilities": [0.0027363125700503588, 0.014792364090681076, 0.003981310874223709, 0.9741946458816528, 0.0027363125700503588, 0.0015591038390994072], "decide_seconds": 0.170635291000508, "probabilities_f64_minus_runtime_max_abs": 4.8077454795070196e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.014824044679406062, "sagan": 0.9762810470164021, "socrates": 0.004247159909519615, "ibn_sina": 0.0012953105883639045, "darwin": 0.002135606119198667, "hypatia": 0.0012168316871096607}, "confidence": 0.9224371465379588, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 178.2656430004863, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "26ba57f0d7982bb838d8721008767af0625af79cf648c0800a20556c8564704e", "compiled_ids_sha256": "944d31185c967912debb5165ee772499c86190aeb63ef728cd1e04b9014126aa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 19.5, 14.0625, 12.875, 13.375, 12.8125], "runtime_probabilities": [0.014824043028056622, 0.976280927658081, 0.004247159231454134, 0.0012953104451298714, 0.0021356060169637203, 0.0012168316170573235], "decide_seconds": 0.17435404299976653, "probabilities_f64_minus_runtime_max_abs": 1.1935832100462562e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.003107011713646269, "ibn_sina": 0.0014676484125010407, "hypatia": 0.0022731383290238985, "einstein": 0.01482272387905428, "sagan": 0.9761940618259075, "darwin": 0.002135415839867047}, "confidence": 0.9216248652094418, "choice": "sagan"}}, "usage": {"input_tokens": 327, "output_tokens": 1}, "latency_ms": 171.64088400022592, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 327, "compiled_prompt_sha256": "f83db599e2edc116912d3740c5169e74befbc33c37d53e395ea2b3a2388f1c99", "compiled_ids_sha256": "043c2fcc4ed7ad12eb2a90c0d51efe2d25c83ad18257b651a77263792833aeb8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.625, 12.875, 13.3125, 15.1875, 19.375, 13.25], "runtime_probabilities": [0.003107012016698718, 0.0014676484279334545, 0.00227313837967813, 0.014822724275290966, 0.9761941432952881, 0.0021354160271584988], "decide_seconds": 0.16751369499979774, "probabilities_f64_minus_runtime_max_abs": 8.146938057507214e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:12Z"} {"id": "c-sagan-01-dessert-03", "uid": "c076", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "prompt_sha256_by_order": ["16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac"], "request_sha256": "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "compiled_prompt_sha256_by_order": ["d1ab3fa14e7acc3c64152e9b9fefb12997ba7cb18617a7d4ba7df9edddd170aa", "a67d01f47b4e062017ef0625a80f0841b27f2ee07b838a1444e1364e594abbec", "ffcbf6fb24313235b7d591c15b4afb758fa9a761ccf6d0770a2a4294ee2bcaf7", "e061af839a84e5f40968432bbf43b1b0645d1b6f93e35f31dfdee2362b65e1b5", "067e4724717953ff29f69852ea64f6abec0a34d07cfeaa8a1edce8084c915f87", "27dec1ce00447b2c51aa4dbbfa98f2d68b019252e4fcb2ce695c4187f86b8845"], "compiled_tokens_by_order": [409, 409, 409, 409, 409, 409], "choice": "sagan", "probs": {"darwin": 0.009773234921964805, "einstein": 0.09272587120627611, "hypatia": 0.007611403010366991, "ibn_sina": 0.00298066824576444, "sagan": 0.879758571201351, "socrates": 0.007150251414276565}, "confidence": 0.7386714928307514, "seconds": 0.17978564399891184, "seconds_all_orders": 1.078713863993471, "seconds_by_order": [0.17778566599736223, 0.17771771700063255, 0.1838288899998588, 0.17887668699768255, 0.17808471399985137, 0.18242018999808352], "argmax_stable": true, "spread": {"darwin": 0.024840400364877703, "einstein": 0.40513664006309963, "hypatia": 0.005544214870372403, "ibn_sina": 0.003475276981139141, "sagan": 0.41957866860552284, "socrates": 0.01019484809809173}, "spread_2dp": {"darwin": 0.03, "einstein": 0.41, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.41999999999999993, "socrates": 0.01}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.009773234921964805, "einstein": 0.09272587120627611, "hypatia": 0.007611403010366991, "ibn_sina": 0.00298066824576444, "sagan": 0.879758571201351, "socrates": 0.007150251414276565}, {"ibn_sina": 0.006455945226903581, "hypatia": 0.003915728717542456, "sagan": 0.5459358330055903, "socrates": 0.006455945226903581, "darwin": 0.012061293571566467, "einstein": 0.4251752542514935}, {"hypatia": 0.003945837088607166, "darwin": 0.003945837088607166, "sagan": 0.9655145016111132, "ibn_sina": 0.0034821890087992414, "einstein": 0.020038614188393856, "socrates": 0.003073021014479453}, {"ibn_sina": 0.006158129104643635, "einstein": 0.04015604601102157, "darwin": 0.02759881990201211, "sagan": 0.9139473945816834, "hypatia": 0.007907194289606952, "socrates": 0.004232416111032219}, {"einstein": 0.23517860720836514, "sagan": 0.7244011084282819, "socrates": 0.013267869112571183, "ibn_sina": 0.004046471456847905, "darwin": 0.019304635649069087, "hypatia": 0.0038013081448647983}, {"socrates": 0.00566834623979142, "ibn_sina": 0.0034380257843003147, "hypatia": 0.009345523015237202, "einstein": 0.0572482690320385, "sagan": 0.8955135984751478, "darwin": 0.02878623745348487}], "prompt_tokens": 409, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.009773234921964805, "einstein": 0.09272587120627611, "hypatia": 0.007611403010366991, "ibn_sina": 0.00298066824576444, "sagan": 0.879758571201351, "socrates": 0.007150251414276565}, "confidence": 0.7386714928307514, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.75585699928342, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "d1ab3fa14e7acc3c64152e9b9fefb12997ba7cb18617a7d4ba7df9edddd170aa", "compiled_ids_sha256": "b4fb76f1da222baeff1f762980a440de6c084e5a94b0a0af875b6114f77bd052", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 17.0, 14.5, 13.5625, 19.25, 14.4375], "runtime_probabilities": [0.009773235768079758, 0.09272586554288864, 0.007611403241753578, 0.002980668330565095, 0.8797585964202881, 0.007150251884013414], "decide_seconds": 0.17272806499750004, "probabilities_f64_minus_runtime_max_abs": 2.5218937049764634e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.006455945226903581, "hypatia": 0.003915728717542456, "sagan": 0.5459358330055903, "socrates": 0.006455945226903581, "darwin": 0.012061293571566467, "einstein": 0.4251752542514935}, "confidence": 0.5344454620313791, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.69158699957188, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "a67d01f47b4e062017ef0625a80f0841b27f2ee07b838a1444e1364e594abbec", "compiled_ids_sha256": "a51e810d1d269d4c0966ebc80dc8ce4fdabfcb95cdc97aeefa02e27a1f643428", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.0625, 19.0, 14.5625, 15.1875, 18.75], "runtime_probabilities": [0.00645594485104084, 0.003915728535503149, 0.5459358096122742, 0.00645594485104084, 0.012061293236911297, 0.4251752495765686], "decide_seconds": 0.17280361400116817, "probabilities_f64_minus_runtime_max_abs": 2.339331617307039e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.003945837088607166, "darwin": 0.003945837088607166, "sagan": 0.9655145016111132, "ibn_sina": 0.0034821890087992414, "einstein": 0.020038614188393856, "socrates": 0.003073021014479453}, "confidence": 0.8920585141497034, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 183.80284999875585, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "ffcbf6fb24313235b7d591c15b4afb758fa9a761ccf6d0770a2a4294ee2bcaf7", "compiled_ids_sha256": "6095298403e624d418b700421225f1067cfe6c176c1249a786e99ae584fc212e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 14.375, 19.875, 14.25, 16.0, 14.125], "runtime_probabilities": [0.00394583772867918, 0.00394583772867918, 0.965514600276947, 0.0034821894951164722, 0.02003861591219902, 0.003073021536692977], "decide_seconds": 0.1789203570006066, "probabilities_f64_minus_runtime_max_abs": 9.866583383644212e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.006158129104643635, "einstein": 0.04015604601102157, "darwin": 0.02759881990201211, "sagan": 0.9139473945816834, "hypatia": 0.007907194289606952, "socrates": 0.004232416111032219}, "confidence": 0.7749894225716296, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 178.850677999435, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "e061af839a84e5f40968432bbf43b1b0645d1b6f93e35f31dfdee2362b65e1b5", "compiled_ids_sha256": "cec607b458d7c156c4c19333a064f264e5ba4ca458982a0cac65ad5d9e6ec02a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 16.375, 16.0, 19.5, 14.75, 14.125], "runtime_probabilities": [0.006158128846436739, 0.0401560440659523, 0.027598818764090538, 0.913947343826294, 0.007907194085419178, 0.004232415929436684], "decide_seconds": 0.1739014060003683, "probabilities_f64_minus_runtime_max_abs": 5.0755389446877075e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.23517860720836514, "sagan": 0.7244011084282819, "socrates": 0.013267869112571183, "ibn_sina": 0.004046471456847905, "darwin": 0.019304635649069087, "hypatia": 0.0038013081448647983}, "confidence": 0.5808674724653786, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 178.05864399997517, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "067e4724717953ff29f69852ea64f6abec0a34d07cfeaa8a1edce8084c915f87", "compiled_ids_sha256": "d256dd8fd200dd22d074739a7cdb22e47a72af6177a7d756491befc1505fa8b9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.0, 19.125, 15.125, 13.9375, 15.5, 13.875], "runtime_probabilities": [0.2351786196231842, 0.7244011163711548, 0.01326787006109953, 0.004046471789479256, 0.019304636865854263, 0.0038013083394616842], "decide_seconds": 0.17315573199812206, "probabilities_f64_minus_runtime_max_abs": 1.241481906166797e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.00566834623979142, "ibn_sina": 0.0034380257843003147, "hypatia": 0.009345523015237202, "einstein": 0.0572482690320385, "sagan": 0.8955135984751478, "darwin": 0.02878623745348487}, "confidence": 0.7448305527551142, "choice": "sagan"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 182.39336000260664, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "27dec1ce00447b2c51aa4dbbfa98f2d68b019252e4fcb2ce695c4187f86b8845", "compiled_ids_sha256": "16e0789b1b4d0bf6f3bd9dcc79956389be010a3f0f11e87d8bb68082525267a9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.8125, 14.8125, 16.625, 19.375, 15.9375], "runtime_probabilities": [0.0056683458387851715, 0.0034380259457975626, 0.009345523081719875, 0.05724826455116272, 0.8955135941505432, 0.02878623828291893], "decide_seconds": 0.17729698000039207, "probabilities_f64_minus_runtime_max_abs": 4.480875781531779e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:13Z"} {"id": "c-sagan-01-dessert-03", "uid": "c077", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "prompt_sha256_by_order": ["6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a"], "request_sha256": "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "compiled_prompt_sha256_by_order": ["5323ed669bcbd837510d737e7f7ff27ae1e5464e1a1530c9bd5f04eaf16e3def", "3b2a4e1f58d86cfc7f81af206b22693e4ba145aa2204a6f8a8d20b76e740db2d", "d32f19300ec41a442fabedb83170a6d3e9541c9fdf4b77a2b95e13086fb906d0", "891cf6a7cd0b6302084f885f9deeadf316145175766fdbedb07875afee58a437", "886564d5a6454a34e2cf4fec815135d2c7987a284381791ba4d8b418d1d15f9f", "75029d7e926a01073aaacf9913cc9a5027c779c262e4160a604938a82b96aa5b"], "compiled_tokens_by_order": [331, 331, 331, 331, 331, 331], "choice": "sagan", "probs": {"darwin": 0.0014915400387910565, "einstein": 0.001587736106547209, "hypatia": 0.0019151753198151367, "ibn_sina": 0.0007499934313121454, "sagan": 0.9920853771525342, "socrates": 0.0021701779510002865}, "confidence": 0.9673421931615234, "seconds": 0.17210431133389648, "seconds_all_orders": 1.0326258680033789, "seconds_by_order": [0.17104718799964758, 0.17312580200086813, 0.17189236199919833, 0.17422667300343164, 0.1707360799991875, 0.17159776300104568], "argmax_stable": true, "spread": {"darwin": 0.001396499078257381, "einstein": 0.002707816651903237, "hypatia": 0.0024727288489728555, "ibn_sina": 0.0010409279269905625, "sagan": 0.006113140317825061, "socrates": 0.00263779719603824}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0014915400387910565, "einstein": 0.001587736106547209, "hypatia": 0.0019151753198151367, "ibn_sina": 0.0007499934313121454, "sagan": 0.9920853771525342, "socrates": 0.0021701779510002865}, {"ibn_sina": 0.0014884093754623322, "hypatia": 0.0010889435613589526, "sagan": 0.9900030426335544, "socrates": 0.0020344144063807103, "darwin": 0.0015844035327810446, "einstein": 0.0038007864904626367}, {"hypatia": 0.0014034048525426892, "darwin": 0.0009061062750259404, "sagan": 0.9936661862535426, "ibn_sina": 0.001092972798126456, "einstein": 0.001238500435441114, "socrates": 0.0016928293853211828}, {"ibn_sina": 0.0017909213583027078, "einstein": 0.002160263516718837, "darwin": 0.002160263516718837, "sagan": 0.9875530459357176, "hypatia": 0.003561672410331808, "socrates": 0.0027738332622102304}, {"einstein": 0.0021630975473785905, "sagan": 0.9888486080690658, "socrates": 0.004041198248421606, "ibn_sina": 0.0009598687180922308, "darwin": 0.0023026053532833215, "hypatia": 0.0016846220637582814}, {"socrates": 0.0014034010523833662, "ibn_sina": 0.0007996338158573444, "hypatia": 0.0018020026210665667, "einstein": 0.0010929698385593993, "sagan": 0.993663495590321, "darwin": 0.001238497081812282}], "prompt_tokens": 331, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0014915400387910565, "einstein": 0.001587736106547209, "hypatia": 0.0019151753198151367, "ibn_sina": 0.0007499934313121454, "sagan": 0.9920853771525342, "socrates": 0.0021701779510002865}, "confidence": 0.9673421931615234, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 171.02025799977127, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "5323ed669bcbd837510d737e7f7ff27ae1e5464e1a1530c9bd5f04eaf16e3def", "compiled_ids_sha256": "9fa1f31f937a41e05c35b71c72dada3a715a34d8f8cc835e0cc5cabd5ccf3784", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.875, 12.9375, 13.125, 12.1875, 19.375, 13.25], "runtime_probabilities": [0.001491539878770709, 0.00158773607108742, 0.0019151752348989248, 0.0007499933708459139, 0.9920852780342102, 0.002170177875086665], "decide_seconds": 0.1669688789988868, "probabilities_f64_minus_runtime_max_abs": 9.911832399289722e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0014884093754623322, "hypatia": 0.0010889435613589526, "sagan": 0.9900030426335544, "socrates": 0.0020344144063807103, "darwin": 0.0015844035327810446, "einstein": 0.0038007864904626367}, "confidence": 0.9603352099437562, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 173.1012730015209, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "3b2a4e1f58d86cfc7f81af206b22693e4ba145aa2204a6f8a8d20b76e740db2d", "compiled_ids_sha256": "a39d341f450e8a28df8e350fd5dae9f3ae4bedb1db746cb33100525299d1b4f9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.125, 12.8125, 19.625, 13.4375, 13.1875, 14.0625], "runtime_probabilities": [0.001488409354351461, 0.0010889435652643442, 0.9900031089782715, 0.0020344143267720938, 0.001584403682500124, 0.0038007867988198996], "decide_seconds": 0.1692557520000264, "probabilities_f64_minus_runtime_max_abs": 6.634471705346812e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0014034048525426892, "darwin": 0.0009061062750259404, "sagan": 0.9936661862535426, "ibn_sina": 0.001092972798126456, "einstein": 0.001238500435441114, "socrates": 0.0016928293853211828}, "confidence": 0.9729725542240591, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 171.86817199763027, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "d32f19300ec41a442fabedb83170a6d3e9541c9fdf4b77a2b95e13086fb906d0", "compiled_ids_sha256": "37ff1b1589d7ddff42626f8c3eb41433e939c949a7974cc697d4218162ac5d5a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 12.75, 19.75, 12.9375, 13.0625, 13.375], "runtime_probabilities": [0.001403404981829226, 0.0009061063174158335, 0.9936662316322327, 0.0010929728159680963, 0.00123850058298558, 0.00169282965362072], "decide_seconds": 0.16786495199994533, "probabilities_f64_minus_runtime_max_abs": 4.5378690027853e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0017909213583027078, "einstein": 0.002160263516718837, "darwin": 0.002160263516718837, "sagan": 0.9875530459357176, "hypatia": 0.003561672410331808, "socrates": 0.0027738332622102304}, "confidence": 0.951654134044374, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 174.20374400171568, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "891cf6a7cd0b6302084f885f9deeadf316145175766fdbedb07875afee58a437", "compiled_ids_sha256": "0f6135a6ba8e2dd16425ebfaf1e9d945e5ccb52a05e28902a3e4ecc79b5deb93", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 13.375, 13.375, 19.5, 13.875, 13.625], "runtime_probabilities": [0.0017909212037920952, 0.002160263480618596, 0.002160263480618596, 0.9875530004501343, 0.0035616722889244556, 0.00277383322827518], "decide_seconds": 0.17034486400007154, "probabilities_f64_minus_runtime_max_abs": 4.548558329986463e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0021630975473785905, "sagan": 0.9888486080690658, "socrates": 0.004041198248421606, "ibn_sina": 0.0009598687180922308, "darwin": 0.0023026053532833215, "hypatia": 0.0016846220637582814}, "confidence": 0.9564407506320033, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 170.71362000206136, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "886564d5a6454a34e2cf4fec815135d2c7987a284381791ba4d8b418d1d15f9f", "compiled_ids_sha256": "1eb2aa9e51b7d6bf68f29de59d25eb9c740ee1bc861706b27b1c403c0f167a22", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 19.625, 14.125, 12.6875, 13.5625, 13.25], "runtime_probabilities": [0.002163097495213151, 0.9888485670089722, 0.0040411981754004955, 0.0009598686010576785, 0.0023026051931083202, 0.0016846221406012774], "decide_seconds": 0.16686119899895857, "probabilities_f64_minus_runtime_max_abs": 4.1060093658451535e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0014034010523833662, "ibn_sina": 0.0007996338158573444, "hypatia": 0.0018020026210665667, "einstein": 0.0010929698385593993, "sagan": 0.993663495590321, "darwin": 0.001238497081812282}, "confidence": 0.9730056828244482, "choice": "sagan"}}, "usage": {"input_tokens": 331, "output_tokens": 1}, "latency_ms": 171.57510299875867, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 331, "compiled_prompt_sha256": "75029d7e926a01073aaacf9913cc9a5027c779c262e4160a604938a82b96aa5b", "compiled_ids_sha256": "acf11c529b163e839cd71470844a2035a26f5f1e36b3a85c914e999481b03def", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.9375, 12.375, 13.1875, 12.6875, 19.5, 12.8125], "runtime_probabilities": [0.001403401023708284, 0.0007996337953954935, 0.0018020024290308356, 0.001092969672754407, 0.9936634302139282, 0.0012384970905259252], "decide_seconds": 0.16766812399873743, "probabilities_f64_minus_runtime_max_abs": 6.537639274384333e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:14Z"} {"id": "c-sagan-01-dessert-03", "uid": "c078", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "prompt_sha256_by_order": ["2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541"], "request_sha256": "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "compiled_prompt_sha256_by_order": ["9185a3cc0cf1abb59a3142933ef89d11e19cf783575408d4218a74be6254ce2d", "629419de84391ebb9e415b1e7674cd3c9f7fc69557a4d8bc4a2fbb9fee405b78", "9f5c19734b49d983625517cb775de8de6568893570aff88022d6f850013da9c2", "3511fd6fe05db9be3342e71a4e01edf6f2892e190e7075a3448d8b10469389f2", "0e23ad47c41478ec22e152aa0e34d025686f509c4fdc87d87f7e2cf6a3be4dcc", "7f109023050a96095c010c7db76a30b2ae9d06086909fda1e15a4b54c3c33587"], "compiled_tokens_by_order": [360, 360, 360, 360, 360, 360], "choice": "sagan", "probs": {"darwin": 0.002451565593386262, "einstein": 0.003147872532585125, "hypatia": 0.002163499042646342, "ibn_sina": 0.0009018805804900735, "sagan": 0.989032149508121, "socrates": 0.002303032742771125}, "confidence": 0.956821864510417, "seconds": 0.17260569099926215, "seconds_all_orders": 1.035634145995573, "seconds_by_order": [0.1711514669987082, 0.17082407899943064, 0.1708330500005104, 0.17564704200049164, 0.17110020799736958, 0.17607829999906244], "argmax_stable": true, "spread": {"darwin": 0.015165524042923846, "einstein": 0.005668178425287761, "hypatia": 0.0023841618462761664, "ibn_sina": 0.0018174074507402772, "sagan": 0.024347285842363564, "socrates": 0.00606630559728956}, "spread_2dp": {"darwin": 0.02, "einstein": 0.01, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.020000000000000018, "socrates": 0.01}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.002451565593386262, "einstein": 0.003147872532585125, "hypatia": 0.002163499042646342, "ibn_sina": 0.0009018805804900735, "sagan": 0.989032149508121, "socrates": 0.002303032742771125}, {"ibn_sina": 0.0016842886521411442, "hypatia": 0.0011575935330059056, "sagan": 0.9886529003072996, "socrates": 0.0023021496336349903, "darwin": 0.0021626694383879673, "einstein": 0.004040398435530534}, {"hypatia": 0.0015883690774536917, "darwin": 0.0014921346599289873, "sagan": 0.9924808844272601, "ibn_sina": 0.00123702308132734, "einstein": 0.0017998579630158326, "socrates": 0.0014017307910140341}, {"ibn_sina": 0.0027192880312303506, "einstein": 0.006128010174830113, "darwin": 0.016657658702852834, "sagan": 0.9681335985848966, "hypatia": 0.003081357026031807, "socrates": 0.003280087480158473}, {"einstein": 0.007468036388303594, "sagan": 0.9781197421056433, "socrates": 0.007468036388303594, "ibn_sina": 0.001565385466647191, "darwin": 0.003997351825435182, "hypatia": 0.0013814478256670875}, {"socrates": 0.00401333462917459, "ibn_sina": 0.0013869713348396797, "hypatia": 0.003541755379282072, "einstein": 0.005485578041782499, "sagan": 0.982030605235639, "darwin": 0.003541755379282072}], "prompt_tokens": 360, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.002451565593386262, "einstein": 0.003147872532585125, "hypatia": 0.002163499042646342, "ibn_sina": 0.0009018805804900735, "sagan": 0.989032149508121, "socrates": 0.002303032742771125}, "confidence": 0.956821864510417, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 171.12324700065074, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "9185a3cc0cf1abb59a3142933ef89d11e19cf783575408d4218a74be6254ce2d", "compiled_ids_sha256": "36b52d662b922ffae58003f7a5d011a210ce57c6309ab056b16d4638019782b0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 13.75, 13.375, 12.5, 19.5, 13.4375], "runtime_probabilities": [0.0024515658151358366, 0.0031478728633373976, 0.002163499128073454, 0.0009018806158564985, 0.9890322089195251, 0.0023030326701700687], "decide_seconds": 0.16670963199794642, "probabilities_f64_minus_runtime_max_abs": 5.9411404129683376e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0016842886521411442, "hypatia": 0.0011575935330059056, "sagan": 0.9886529003072996, "socrates": 0.0023021496336349903, "darwin": 0.0021626694383879673, "einstein": 0.004040398435530534}, "confidence": 0.955692496150307, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 170.79895999995642, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "629419de84391ebb9e415b1e7674cd3c9f7fc69557a4d8bc4a2fbb9fee405b78", "compiled_ids_sha256": "83449f2780bb31841eccaca7e90f626abb698216497b6570f7e86111392cea54", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 13.0, 19.75, 13.6875, 13.625, 14.25], "runtime_probabilities": [0.001684288727119565, 0.001157593447715044, 0.9886528849601746, 0.0023021495435386896, 0.002162669552490115, 0.004040398634970188], "decide_seconds": 0.16665615199963213, "probabilities_f64_minus_runtime_max_abs": 1.53471250152748e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0015883690774536917, "darwin": 0.0014921346599289873, "sagan": 0.9924808844272601, "ibn_sina": 0.00123702308132734, "einstein": 0.0017998579630158326, "socrates": 0.0014017307910140341}, "confidence": 0.9685758147866645, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 170.80836000241106, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "9f5c19734b49d983625517cb775de8de6568893570aff88022d6f850013da9c2", "compiled_ids_sha256": "aa0d88f3004823454d8f6f25556abab2cd36241475eb4b7b31805101fc8aff49", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.4375, 13.375, 19.875, 13.1875, 13.5625, 13.3125], "runtime_probabilities": [0.0015883691376075149, 0.001492134528234601, 0.9924808740615845, 0.0012370231561362743, 0.0017998578259721398, 0.0014017308130860329], "decide_seconds": 0.16660792200127617, "probabilities_f64_minus_runtime_max_abs": 1.0365675651158313e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0027192880312303506, "einstein": 0.006128010174830113, "darwin": 0.016657658702852834, "sagan": 0.9681335985848966, "hypatia": 0.003081357026031807, "socrates": 0.003280087480158473}, "confidence": 0.8976262735434417, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 175.6225130011444, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "3511fd6fe05db9be3342e71a4e01edf6f2892e190e7075a3448d8b10469389f2", "compiled_ids_sha256": "a041de6bb22a3abcce3cd5cb9274a1b041ff0956442c449457c627c257a87c8e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 14.5625, 15.5625, 19.625, 13.875, 13.9375], "runtime_probabilities": [0.002719287993386388, 0.006128009874373674, 0.01665765792131424, 0.9681335687637329, 0.0030813568737357855, 0.003280087374150753], "decide_seconds": 0.1714674249997188, "probabilities_f64_minus_runtime_max_abs": 2.982116364957932e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.007468036388303594, "sagan": 0.9781197421056433, "socrates": 0.007468036388303594, "ibn_sina": 0.001565385466647191, "darwin": 0.003997351825435182, "hypatia": 0.0013814478256670875}, "confidence": 0.9240607385024334, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 171.07454800134292, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "0e23ad47c41478ec22e152aa0e34d025686f509c4fdc87d87f7e2cf6a3be4dcc", "compiled_ids_sha256": "9a513e31a566b29634c7a06654ed5f0715fd0ce6ca53c259a9e8fb5538f73553", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 19.625, 14.75, 13.1875, 14.125, 13.0625], "runtime_probabilities": [0.007468035910278559, 0.9781197309494019, 0.007468035910278559, 0.0015653854934498668, 0.003997351974248886, 0.001381447771564126], "decide_seconds": 0.16682522999690264, "probabilities_f64_minus_runtime_max_abs": 1.1156241486176555e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.00401333462917459, "ibn_sina": 0.0013869713348396797, "hypatia": 0.003541755379282072, "einstein": 0.005485578041782499, "sagan": 0.982030605235639, "darwin": 0.003541755379282072}, "confidence": 0.9343609837930562, "choice": "sagan"}}, "usage": {"input_tokens": 360, "output_tokens": 1}, "latency_ms": 176.05360000015935, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 360, "compiled_prompt_sha256": "7f109023050a96095c010c7db76a30b2ae9d06086909fda1e15a4b54c3c33587", "compiled_ids_sha256": "83290ee5bf70a6206a37c4f9a8deb40adc21e5935919d4b8275ac78cbe29490e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 12.9375, 13.875, 14.3125, 19.5, 13.875], "runtime_probabilities": [0.00401333486661315, 0.0013869714457541704, 0.00354175572283566, 0.005485578440129757, 0.9820306301116943, 0.00354175572283566], "decide_seconds": 0.1718625619978411, "probabilities_f64_minus_runtime_max_abs": 2.487605532675019e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:15Z"} {"id": "c-sagan-01-dessert-03", "uid": "c079", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "prompt_sha256_by_order": ["fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520"], "request_sha256": "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "compiled_prompt_sha256_by_order": ["4c9fb986381beb2b96309a2afbd1b2af9fe10613bf4378bdb77d7479f8c41680", "4633d390e8d44b5e0602c3870a0f99da84a4ae306173ae0a3145e5d367fe328c", "5d26fe24f8b07848c7fa8024319fe88766db722129d7d7bffc5883e3ab2aed19", "81e4846b544328a97f2990eae522b97b7e247176c221d1e89c48ba2f68f56510", "be489bc7e710f529e7158e9eb3d0196ec5bec446b9164b7051a0bd158fda5456", "288eaf677a09cb6634d10a4c1ac5a8676a716d57b3249972b2fee3ee7243bf40"], "compiled_tokens_by_order": [397, 397, 397, 397, 397, 397], "choice": "sagan", "probs": {"darwin": 0.010879731236757558, "einstein": 0.007477522635150714, "hypatia": 0.02778236699274203, "ibn_sina": 0.004260563969045589, "sagan": 0.9200256394469077, "socrates": 0.02957417571939632}, "confidence": 0.7826615810103446, "seconds": 0.17860628000016732, "seconds_all_orders": 1.071637680001004, "seconds_by_order": [0.17742555899894796, 0.17821610299870372, 0.18245924100119737, 0.1768452830001479, 0.17927115500060609, 0.17742033900140086], "argmax_stable": true, "spread": {"darwin": 0.06523926570833097, "einstein": 0.04948781789662847, "hypatia": 0.21436533777648284, "ibn_sina": 0.0158681774428363, "sagan": 0.6034393017300252, "socrates": 0.2717514672646104}, "spread_2dp": {"darwin": 0.060000000000000005, "einstein": 0.04, "hypatia": 0.22, "ibn_sina": 0.02, "sagan": 0.6, "socrates": 0.27}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.010879731236757558, "einstein": 0.007477522635150714, "hypatia": 0.02778236699274203, "ibn_sina": 0.004260563969045589, "sagan": 0.9200256394469077, "socrates": 0.02957417571939632}, {"ibn_sina": 0.0051746870394258635, "hypatia": 0.003138606343829383, "sagan": 0.9263747606042874, "socrates": 0.007529125215700709, "darwin": 0.011661337256287513, "einstein": 0.04612148354046925}, {"hypatia": 0.00690025809849284, "darwin": 0.005373926410501027, "sagan": 0.962042680862356, "ibn_sina": 0.0039316485507316775, "einstein": 0.008860106780170232, "socrates": 0.012891379297748203}, {"ibn_sina": 0.006843412810771841, "einstein": 0.009353836276877765, "darwin": 0.014487505034091377, "sagan": 0.8963101979629099, "hypatia": 0.022438686748415256, "socrates": 0.05056636116693382}, {"einstein": 0.05499360931731791, "sagan": 0.3586033791323308, "socrates": 0.2792805924803111, "ibn_sina": 0.01900528283089601, "darwin": 0.07061319211883199, "hypatia": 0.2175039441203122}, {"socrates": 0.00966296487313306, "ibn_sina": 0.003137105388059711, "hypatia": 0.04609942715644326, "einstein": 0.00550579142068944, "sagan": 0.9259317462885414, "darwin": 0.00966296487313306}], "prompt_tokens": 397, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.010879731236757558, "einstein": 0.007477522635150714, "hypatia": 0.02778236699274203, "ibn_sina": 0.004260563969045589, "sagan": 0.9200256394469077, "socrates": 0.02957417571939632}, "confidence": 0.7826615810103446, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 177.396259001398, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "4c9fb986381beb2b96309a2afbd1b2af9fe10613bf4378bdb77d7479f8c41680", "compiled_ids_sha256": "86437e7858dab9217d982273312aa4f196d2a66f6de0c27248ed494a7c977df0", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 14.5625, 15.875, 14.0, 19.375, 15.9375], "runtime_probabilities": [0.010879730805754662, 0.007477523293346167, 0.027782367542386055, 0.004260564222931862, 0.920025646686554, 0.029574178159236908], "decide_seconds": 0.1726373359997524, "probabilities_f64_minus_runtime_max_abs": 7.239646215850826e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0051746870394258635, "hypatia": 0.003138606343829383, "sagan": 0.9263747606042874, "socrates": 0.007529125215700709, "darwin": 0.011661337256287513, "einstein": 0.04612148354046925}, "confidence": 0.8064541296394734, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 178.19032300030813, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "4633d390e8d44b5e0602c3870a0f99da84a4ae306173ae0a3145e5d367fe328c", "compiled_ids_sha256": "29bbb1050c1a409fbef1f2270dbc3cd5e252f9a1569e55ceb84c7330bfb392fa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.0625, 19.75, 14.9375, 15.375, 16.75], "runtime_probabilities": [0.005174687597900629, 0.0031386064365506172, 0.9263747930526733, 0.007529126014560461, 0.011661337688565254, 0.04612148180603981], "decide_seconds": 0.17371628700129804, "probabilities_f64_minus_runtime_max_abs": 3.244838597815658e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.00690025809849284, "darwin": 0.005373926410501027, "sagan": 0.962042680862356, "ibn_sina": 0.0039316485507316775, "einstein": 0.008860106780170232, "socrates": 0.012891379297748203}, "confidence": 0.87755408877915, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 182.4340109997138, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "5d26fe24f8b07848c7fa8024319fe88766db722129d7d7bffc5883e3ab2aed19", "compiled_ids_sha256": "4813f1dcd6bb58eb899d39cd9b012a9ff146d33efc73347b7b9bed3667ce65a9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 14.5625, 19.75, 14.25, 15.0625, 15.4375], "runtime_probabilities": [0.006900257430970669, 0.005373926367610693, 0.9620426893234253, 0.003931648563593626, 0.008860106579959393, 0.012891379185020924], "decide_seconds": 0.1778052360023139, "probabilities_f64_minus_runtime_max_abs": 8.46106928964474e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.006843412810771841, "einstein": 0.009353836276877765, "darwin": 0.014487505034091377, "sagan": 0.8963101979629099, "hypatia": 0.022438686748415256, "socrates": 0.05056636116693382}, "confidence": 0.7357964891832706, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 176.82057400088524, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "81e4846b544328a97f2990eae522b97b7e247176c221d1e89c48ba2f68f56510", "compiled_ids_sha256": "daef1278477a90357311568da60bcdc7305a2eb22a12173c7047a73b5841b367", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 14.9375, 15.375, 19.5, 15.8125, 16.625], "runtime_probabilities": [0.006843412294983864, 0.009353836067020893, 0.01448750402778387, 0.8963101506233215, 0.022438686341047287, 0.05056636035442352], "decide_seconds": 0.17227678800190915, "probabilities_f64_minus_runtime_max_abs": 4.7339588360451046e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.05499360931731791, "sagan": 0.3586033791323308, "socrates": 0.2792805924803111, "ibn_sina": 0.01900528283089601, "darwin": 0.07061319211883199, "hypatia": 0.2175039441203122}, "confidence": 0.17522529487338656, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 179.24652500005323, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "be489bc7e710f529e7158e9eb3d0196ec5bec446b9164b7051a0bd158fda5456", "compiled_ids_sha256": "dbf17730bb89eba96c246ae04ed042811507045da046d7835c92daec8dbba1a6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 18.375, 18.125, 15.4375, 16.75, 17.875], "runtime_probabilities": [0.054993610829114914, 0.3586033880710602, 0.2792806029319763, 0.01900528371334076, 0.07061319053173065, 0.21750394999980927], "decide_seconds": 0.17476862900002743, "probabilities_f64_minus_runtime_max_abs": 1.0451665199973093e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.00966296487313306, "ibn_sina": 0.003137105388059711, "hypatia": 0.04609942715644326, "einstein": 0.00550579142068944, "sagan": 0.9259317462885414, "darwin": 0.00966296487313306}, "confidence": 0.8049475585151381, "choice": "sagan"}}, "usage": {"input_tokens": 397, "output_tokens": 1}, "latency_ms": 177.39448900101706, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 397, "compiled_prompt_sha256": "288eaf677a09cb6634d10a4c1ac5a8676a716d57b3249972b2fee3ee7243bf40", "compiled_ids_sha256": "abeb2e23820f04985eda3cf3420b0e1d784fd5cdfed661960f8f810b7d351c10", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 13.6875, 16.375, 14.25, 19.375, 14.8125], "runtime_probabilities": [0.009662965312600136, 0.0031371053773909807, 0.04609942436218262, 0.005505790933966637, 0.9259316921234131, 0.009662965312600136], "decide_seconds": 0.17282393400091678, "probabilities_f64_minus_runtime_max_abs": 5.416512827149944e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:16Z"} {"id": "c-sagan-01-main-01", "uid": "c080", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "prompt_sha256_by_order": ["140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631"], "request_sha256": "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "compiled_prompt_sha256_by_order": ["9039ec9ebd536e8ff815ec4d3af1f3487d4b1b3ac68f56b8c74d168136f644a8", "d6f6d0e89e0ff3c08e90886efdbf4fc85b17286e73cab77a600bf5a29c8c244a", "634404df4f0f7dd3981cb1ed7a904e2c0c257d58f839af054cc97539723d3ee5", "bc6965f7530f96e3349a0d9519fd5cc3804aa52c359cc414420aafe2f7456b0a", "e9f387a197c782c1116fc9818b4fcee6083b269a08ed87d757dccea28b386fb2", "acec5b43c52adb11448d98ec15993c4c089d7c3bbe6449e7af85b598b61ebfd3"], "compiled_tokens_by_order": [333, 333, 333, 333, 333, 333], "choice": "sagan", "probs": {"darwin": 0.01564848206682648, "einstein": 0.13947340246465273, "hypatia": 0.017732053246868402, "ibn_sina": 0.005756754837925464, "sagan": 0.4868100080457417, "socrates": 0.3345792993379852}, "confidence": 0.3538390322308569, "seconds": 0.17259366433427203, "seconds_all_orders": 1.0355619860056322, "seconds_by_order": [0.17963298200265854, 0.17224914799953694, 0.17103118800150696, 0.17073352000079467, 0.17083946000275319, 0.1710756879983819], "argmax_stable": false, "spread": {"darwin": 0.013802719634661337, "einstein": 0.8122232028215475, "hypatia": 0.06758419708290771, "ibn_sina": 0.008962186016685685, "sagan": 0.8713216896013216, "socrates": 0.37406069512910944}, "spread_2dp": {"darwin": 0.01, "einstein": 0.8099999999999999, "hypatia": 0.060000000000000005, "ibn_sina": 0.01, "sagan": 0.87, "socrates": 0.38}, "choices_by_order": ["sagan", "einstein", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.01564848206682648, "einstein": 0.13947340246465273, "hypatia": 0.017732053246868402, "ibn_sina": 0.005756754837925464, "sagan": 0.4868100080457417, "socrates": 0.3345792993379852}, {"ibn_sina": 0.00790203335135481, "hypatia": 0.0051019361890862905, "sagan": 0.04840582281600845, "socrates": 0.07043005661482214, "darwin": 0.010146411666653792, "einstein": 0.8580137393620745}, {"hypatia": 0.008276218199074344, "darwin": 0.009983024742531906, "sagan": 0.42448901368585484, "ibn_sina": 0.005688156037158525, "einstein": 0.17695334757642345, "socrates": 0.37461023975895685}, {"ibn_sina": 0.013221369383685028, "einstein": 0.1254408601073219, "darwin": 0.012420327107315868, "sagan": 0.43783162265354697, "hypatia": 0.02470076990278771, "socrates": 0.3863850508453426}, {"einstein": 0.28747923378316637, "sagan": 0.47397312762288124, "socrates": 0.13579557464678255, "ibn_sina": 0.009241004827530574, "darwin": 0.020824925847645347, "hypatia": 0.072686133271994}, {"socrates": 0.012324355716233129, "ibn_sina": 0.004259183366999345, "hypatia": 0.01087620574592649, "einstein": 0.045790536540527046, "sagan": 0.9197275124173301, "darwin": 0.00702220621298401}], "prompt_tokens": 333, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.01564848206682648, "einstein": 0.13947340246465273, "hypatia": 0.017732053246868402, "ibn_sina": 0.005756754837925464, "sagan": 0.4868100080457417, "socrates": 0.3345792993379852}, "confidence": 0.3538390322308569, "choice": "sagan"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 179.60512199715595, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "9039ec9ebd536e8ff815ec4d3af1f3487d4b1b3ac68f56b8c74d168136f644a8", "compiled_ids_sha256": "d5b1938a23c4d5daa57a1a68304eaa2e03f3dfd41e38df0bac9ead15b4c8f182", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 17.5, 15.4375, 14.3125, 18.75, 18.375], "runtime_probabilities": [0.015648482367396355, 0.13947340846061707, 0.017732053995132446, 0.005756754893809557, 0.48681002855300903, 0.33457931876182556], "decide_seconds": 0.17537625399927492, "probabilities_f64_minus_runtime_max_abs": 2.050726732871766e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.00790203335135481, "hypatia": 0.0051019361890862905, "sagan": 0.04840582281600845, "socrates": 0.07043005661482214, "darwin": 0.010146411666653792, "einstein": 0.8580137393620745}, "confidence": 0.6781990191821418, "choice": "einstein"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 172.22359900188167, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "d6f6d0e89e0ff3c08e90886efdbf4fc85b17286e73cab77a600bf5a29c8c244a", "compiled_ids_sha256": "de128e35481af2b6834947a1950201ca68fafdef653a21826bc449d4c3b2b234", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.125, 16.375, 16.75, 14.8125, 19.25], "runtime_probabilities": [0.007902032695710659, 0.005101935938000679, 0.04840582609176636, 0.07043005526065826, 0.010146411135792732, 0.8580137491226196], "decide_seconds": 0.16793595199851552, "probabilities_f64_minus_runtime_max_abs": 9.760545149894995e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.008276218199074344, "darwin": 0.009983024742531906, "sagan": 0.42448901368585484, "ibn_sina": 0.005688156037158525, "einstein": 0.17695334757642345, "socrates": 0.37461023975895685}, "confidence": 0.3564517124533705, "choice": "sagan"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 171.00603799917735, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "634404df4f0f7dd3981cb1ed7a904e2c0c257d58f839af054cc97539723d3ee5", "compiled_ids_sha256": "09a03a9a11ea7ea49304c88d259b543b72940095202b57c28b0db96b7a417e26", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 15.25, 19.0, 14.6875, 18.125, 18.875], "runtime_probabilities": [0.008276219479739666, 0.00998302549123764, 0.42448902130126953, 0.005688156466931105, 0.17695336043834686, 0.3746102452278137], "decide_seconds": 0.16686977999779629, "probabilities_f64_minus_runtime_max_abs": 1.286192341520831e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013221369383685028, "einstein": 0.1254408601073219, "darwin": 0.012420327107315868, "sagan": 0.43783162265354697, "hypatia": 0.02470076990278771, "socrates": 0.3863850508453426}, "confidence": 0.33442126899015545, "choice": "sagan"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 170.70963000151096, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "bc6965f7530f96e3349a0d9519fd5cc3804aa52c359cc414420aafe2f7456b0a", "compiled_ids_sha256": "b1dea77649d622bea84c135bb44656e0a4ac92d2bf541d61138ef152418d50d1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 17.5, 15.1875, 18.75, 15.875, 18.625], "runtime_probabilities": [0.013221370056271553, 0.12544086575508118, 0.01242032740265131, 0.4378316402435303, 0.024700772017240524, 0.386385053396225], "decide_seconds": 0.16679230099907727, "probabilities_f64_minus_runtime_max_abs": 1.7589983303256673e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.28747923378316637, "sagan": 0.47397312762288124, "socrates": 0.13579557464678255, "ibn_sina": 0.009241004827530574, "darwin": 0.020824925847645347, "hypatia": 0.072686133271994}, "confidence": 0.2756619351930195, "choice": "sagan"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 170.815550001862, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "e9f387a197c782c1116fc9818b4fcee6083b269a08ed87d757dccea28b386fb2", "compiled_ids_sha256": "ca7308b553e834c9c5addc2626505d0add7bd63a03330e9ea36806d63f6dd168", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.25, 18.75, 17.5, 14.8125, 15.625, 16.875], "runtime_probabilities": [0.2874792516231537, 0.4739731550216675, 0.13579557836055756, 0.009241005405783653, 0.02082492597401142, 0.07268613576889038], "decide_seconds": 0.16685101999973995, "probabilities_f64_minus_runtime_max_abs": 2.7398786239807293e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.012324355716233129, "ibn_sina": 0.004259183366999345, "hypatia": 0.01087620574592649, "einstein": 0.045790536540527046, "sagan": 0.9197275124173301, "darwin": 0.00702220621298401}, "confidence": 0.7881480388327089, "choice": "sagan"}}, "usage": {"input_tokens": 333, "output_tokens": 1}, "latency_ms": 171.05209800138255, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 333, "compiled_prompt_sha256": "acec5b43c52adb11448d98ec15993c4c089d7c3bbe6449e7af85b598b61ebfd3", "compiled_ids_sha256": "cf87871ca9b9eaca3f396be607278ce4ece6257cf2ffa8793f97646a1fbcdfe9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.0, 14.9375, 16.375, 19.375, 14.5], "runtime_probabilities": [0.012324356473982334, 0.004259184002876282, 0.010876205749809742, 0.04579053819179535, 0.9197275638580322, 0.0070222062058746815], "decide_seconds": 0.16711590900013107, "probabilities_f64_minus_runtime_max_abs": 5.1440702142357964e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:17Z"} {"id": "c-sagan-01-main-02", "uid": "c081", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "prompt_sha256_by_order": ["365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3"], "request_sha256": "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "compiled_prompt_sha256_by_order": ["e899284aeb4ab20e1a02f56556620f4fbd0c64a5bfca3086c98d666d6ed9d79b", "84be2b1a1fa4ad7b85a75275ba7652beae8de0a0ad537f6a84c2359c868bc409", "da020ac09be783b39bd67ca45116b423d2f97a5b0a0d1c902937ce9be596e7a9", "caf4e2a091ac5fd1feeadb248dbe1d0d2a062523f10f2d26ea30bad9466c54f8", "791f873348b9933d8fff5f837bbf63f77beb0db4d82fed4393b6e99a8d058dcd", "49981d7b5cb7600ad39ead1b3834a2e527bcbf1f27e0eb00c993b68564a5bb4a"], "compiled_tokens_by_order": [366, 366, 366, 366, 366, 366], "choice": "sagan", "probs": {"darwin": 0.00677795107451132, "einstein": 0.009264360692061773, "hypatia": 0.026807316523894235, "ibn_sina": 0.012662879694629162, "sagan": 0.8877364024884782, "socrates": 0.056751089526425325}, "confidence": 0.7220051393420388, "seconds": 0.1710976658329552, "seconds_all_orders": 1.0265859949977312, "seconds_by_order": [0.1715400339999178, 0.17108298799939803, 0.17077707999851555, 0.17112441799690714, 0.17114961700281128, 0.17091185800018138], "argmax_stable": false, "spread": {"darwin": 0.03658589377923112, "einstein": 0.07837454067496064, "hypatia": 0.06387611742784732, "ibn_sina": 0.05840569511460505, "sagan": 0.5448613959827329, "socrates": 0.4316160178887467}, "spread_2dp": {"darwin": 0.04, "einstein": 0.08, "hypatia": 0.07, "ibn_sina": 0.060000000000000005, "sagan": 0.55, "socrates": 0.43}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.00677795107451132, "einstein": 0.009264360692061773, "hypatia": 0.026807316523894235, "ibn_sina": 0.012662879694629162, "sagan": 0.8877364024884782, "socrates": 0.056751089526425325}, {"ibn_sina": 0.0091103968020902, "hypatia": 0.013255549130806357, "sagan": 0.8200917851336076, "socrates": 0.05940724966086408, "darwin": 0.011697981049994536, "einstein": 0.08643703822263729}, {"hypatia": 0.01262963768467355, "darwin": 0.00436468599222085, "sagan": 0.8317619225311803, "ibn_sina": 0.041410987702845826, "einstein": 0.022165704434207595, "socrates": 0.08766706165487187}, {"ibn_sina": 0.009136006624004878, "einstein": 0.00806249754767665, "darwin": 0.014150123867780895, "sagan": 0.8223971078354841, "hypatia": 0.05957424667727357, "socrates": 0.08668001744777995}, {"einstein": 0.03189234359323482, "sagan": 0.3428750065057453, "socrates": 0.44026022310035173, "ibn_sina": 0.06751609191669525, "darwin": 0.040950579771451966, "hypatia": 0.07650575511252088}, {"socrates": 0.008644205211605055, "ibn_sina": 0.012577244369083267, "hypatia": 0.06387251723921844, "einstein": 0.011099379198765526, "sagan": 0.8817329026796404, "darwin": 0.022073751301687282}], "prompt_tokens": 366, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.00677795107451132, "einstein": 0.009264360692061773, "hypatia": 0.026807316523894235, "ibn_sina": 0.012662879694629162, "sagan": 0.8877364024884782, "socrates": 0.056751089526425325}, "confidence": 0.7220051393420388, "choice": "sagan"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 171.51230500167003, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "e899284aeb4ab20e1a02f56556620f4fbd0c64a5bfca3086c98d666d6ed9d79b", "compiled_ids_sha256": "baf3a4db52592f705e8840880143e8aa8842446449b21f0d838487828fea0bfd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 14.9375, 16.0, 15.25, 19.5, 16.75], "runtime_probabilities": [0.0067779505625367165, 0.00926436111330986, 0.02680731564760208, 0.012662879191339016, 0.8877363801002502, 0.05675108730792999], "decide_seconds": 0.1670205789996544, "probabilities_f64_minus_runtime_max_abs": 2.238822793820816e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0091103968020902, "hypatia": 0.013255549130806357, "sagan": 0.8200917851336076, "socrates": 0.05940724966086408, "darwin": 0.011697981049994536, "einstein": 0.08643703822263729}, "confidence": 0.6125826399040382, "choice": "sagan"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 171.05680800159462, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "84be2b1a1fa4ad7b85a75275ba7652beae8de0a0ad537f6a84c2359c868bc409", "compiled_ids_sha256": "f987da62115aacea72c40420bceff1f947b1c69d6be7e988b2df76c44a519863", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 15.25, 19.375, 16.75, 15.125, 17.125], "runtime_probabilities": [0.009110397659242153, 0.013255548663437366, 0.8200917840003967, 0.059407249093055725, 0.0116979805752635, 0.08643703162670135], "decide_seconds": 0.16678560099899187, "probabilities_f64_minus_runtime_max_abs": 6.595935936326569e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.01262963768467355, "darwin": 0.00436468599222085, "sagan": 0.8317619225311803, "ibn_sina": 0.041410987702845826, "einstein": 0.022165704434207595, "socrates": 0.08766706165487187}, "confidence": 0.6306174009680963, "choice": "sagan"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 170.75250999914715, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "da020ac09be783b39bd67ca45116b423d2f97a5b0a0d1c902937ce9be596e7a9", "compiled_ids_sha256": "a5924d483e2892273033c75267402039f6edffc41769696b4dcb3c95911bd209", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.5, 19.75, 16.75, 16.125, 17.5], "runtime_probabilities": [0.012629637494683266, 0.004364686086773872, 0.8317619562149048, 0.04141099005937576, 0.022165704518556595, 0.0876670628786087], "decide_seconds": 0.16657403200224508, "probabilities_f64_minus_runtime_max_abs": 3.3683724476318844e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.009136006624004878, "einstein": 0.00806249754767665, "darwin": 0.014150123867780895, "sagan": 0.8223971078354841, "hypatia": 0.05957424667727357, "socrates": 0.08668001744777995}, "confidence": 0.6189051551194831, "choice": "sagan"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 171.10025800138828, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "caf4e2a091ac5fd1feeadb248dbe1d0d2a062523f10f2d26ea30bad9466c54f8", "compiled_ids_sha256": "45832ddbd3efe171a06ae499255398d2b2e0b466e7f9f4db14161e02889178ae", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.875, 15.4375, 19.5, 16.875, 17.25], "runtime_probabilities": [0.00913600716739893, 0.008062497712671757, 0.014150123111903667, 0.8223970532417297, 0.059574246406555176, 0.08668000996112823], "decide_seconds": 0.16690506899976754, "probabilities_f64_minus_runtime_max_abs": 5.459375440608483e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03189234359323482, "sagan": 0.3428750065057453, "socrates": 0.44026022310035173, "ibn_sina": 0.06751609191669525, "darwin": 0.040950579771451966, "hypatia": 0.07650575511252088}, "confidence": 0.24791197803049447, "choice": "socrates"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 171.12520799855702, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "791f873348b9933d8fff5f837bbf63f77beb0db4d82fed4393b6e99a8d058dcd", "compiled_ids_sha256": "f8442972e84cb078f152f8cae1731c470b3ea5c422e1d6f83df994bdc447c159", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 18.375, 18.625, 16.75, 16.25, 16.875], "runtime_probabilities": [0.03189234435558319, 0.34287500381469727, 0.4402602016925812, 0.06751609593629837, 0.040950577706098557, 0.07650575041770935], "decide_seconds": 0.16683508999994956, "probabilities_f64_minus_runtime_max_abs": 2.1407770556258754e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.008644205211605055, "ibn_sina": 0.012577244369083267, "hypatia": 0.06387251723921844, "einstein": 0.011099379198765526, "sagan": 0.8817329026796404, "darwin": 0.022073751301687282}, "confidence": 0.71150093163201, "choice": "sagan"}}, "usage": {"input_tokens": 366, "output_tokens": 1}, "latency_ms": 170.8876499978942, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 366, "compiled_prompt_sha256": "49981d7b5cb7600ad39ead1b3834a2e527bcbf1f27e0eb00c993b68564a5bb4a", "compiled_ids_sha256": "805e50fd5896bf2ddfaa93d190dd564ed9384895196e475151b3dd24a10df0e4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 15.125, 16.75, 15.0, 19.375, 15.6875], "runtime_probabilities": [0.008644205518066883, 0.012577243149280548, 0.06387251615524292, 0.011099378578364849, 0.8817328810691833, 0.02207375131547451], "decide_seconds": 0.16671754099661484, "probabilities_f64_minus_runtime_max_abs": 2.1610457090481816e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:18Z"} {"id": "c-sagan-01-main-02", "uid": "c082", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "prompt_sha256_by_order": ["dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc"], "request_sha256": "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "compiled_prompt_sha256_by_order": ["6b3262c0b05359df640a75374ab1cb9ee093c35d018b01ad37614f989d8ad180", "e58f881f1f8ad60e90fb06493132e99ff4ff5bdcd62b203c274caadf5f8e7bd4", "51144324b9565a106912a27b203efc22e0622c9d2a6c6bf22976aed0d6f4c980", "9861ffca80b3747e09e69887723701d4efc811d7e5b8ba2e3cca712dcc6e7fd9", "a4cdd1d51e4bd6a3833600e0339a5be23818ee80cc31a34409d53b098778e816", "d6853c19ae77d785ff4d0e26d079aa1237f28072b8649be382018b78c39bc252"], "compiled_tokens_by_order": [316, 316, 316, 316, 316, 316], "choice": "einstein", "probs": {"darwin": 0.009115720331400274, "einstein": 0.9298287524491742, "hypatia": 0.01812877458528883, "ibn_sina": 0.006669203463559101, "sagan": 0.01812877458528883, "socrates": 0.01812877458528883}, "confidence": 0.7979693136376625, "seconds": 0.16552701500040712, "seconds_all_orders": 0.9931620900024427, "seconds_by_order": [0.16604949600150576, 0.16556631000275956, 0.16533458099729614, 0.1655374900001334, 0.16546524099976523, 0.1652089720009826], "argmax_stable": true, "spread": {"darwin": 0.029225246558190893, "einstein": 0.08507814706772976, "hypatia": 0.0295287934138111, "ibn_sina": 0.022236572695234665, "sagan": 0.037702744277848914, "socrates": 0.04710278178514244}, "spread_2dp": {"darwin": 0.03, "einstein": 0.09000000000000008, "hypatia": 0.030000000000000002, "ibn_sina": 0.019999999999999997, "sagan": 0.03, "socrates": 0.039999999999999994}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.009115720331400274, "einstein": 0.9298287524491742, "hypatia": 0.01812877458528883, "ibn_sina": 0.006669203463559101, "sagan": 0.01812877458528883, "socrates": 0.01812877458528883}, {"ibn_sina": 0.02114051093982601, "hypatia": 0.023955337268506165, "sagan": 0.011315700085272613, "socrates": 0.027144953368501853, "darwin": 0.017526099142327067, "einstein": 0.8989173991955662}, {"hypatia": 0.02106277751937444, "darwin": 0.006423784204603875, "sagan": 0.01359915126786264, "ibn_sina": 0.012775220344206605, "einstein": 0.8956120900503995, "socrates": 0.050526976613552946}, {"ibn_sina": 0.02030965581355217, "einstein": 0.8635885402411037, "darwin": 0.0035292889808422184, "sagan": 0.007018829237184819, "hypatia": 0.04299554169468765, "socrates": 0.06255814403262944}, {"einstein": 0.8447506053814444, "sagan": 0.008815763644651052, "socrates": 0.037115751276977756, "ibn_sina": 0.028905776158793767, "darwin": 0.03275453553903311, "hypatia": 0.04765756799909993}, {"socrates": 0.015455362247486998, "ibn_sina": 0.011307384571293383, "hypatia": 0.022487419379907853, "einstein": 0.8982568160992617, "sagan": 0.04472157351503373, "darwin": 0.007771444187016396}], "prompt_tokens": 316, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.009115720331400274, "einstein": 0.9298287524491742, "hypatia": 0.01812877458528883, "ibn_sina": 0.006669203463559101, "sagan": 0.01812877458528883, "socrates": 0.01812877458528883}, "confidence": 0.7979693136376625, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 166.0231459973147, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "6b3262c0b05359df640a75374ab1cb9ee093c35d018b01ad37614f989d8ad180", "compiled_ids_sha256": "9cf18e6fda3d59410ee91dbb91dadbf48ceba6733de0f541fd3d4d6902aa5f4c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 19.5, 15.5625, 14.5625, 15.5625, 15.5625], "runtime_probabilities": [0.009115721099078655, 0.9298287630081177, 0.01812877506017685, 0.006669202819466591, 0.01812877506017685, 0.01812877506017685], "decide_seconds": 0.16210292700270656, "probabilities_f64_minus_runtime_max_abs": 1.0558943497329665e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02114051093982601, "hypatia": 0.023955337268506165, "sagan": 0.011315700085272613, "socrates": 0.027144953368501853, "darwin": 0.017526099142327067, "einstein": 0.8989173991955662}, "confidence": 0.728645673986012, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 165.54167999856872, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "e58f881f1f8ad60e90fb06493132e99ff4ff5bdcd62b203c274caadf5f8e7bd4", "compiled_ids_sha256": "3fd853485a7de4c9711c7e80a7f456549e61f5bff87ad19f7d61cf288cd8ea90", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.625, 15.75, 15.0, 15.875, 15.4375, 19.375], "runtime_probabilities": [0.021140510216355324, 0.023955335840582848, 0.01131569966673851, 0.02714495360851288, 0.01752609945833683, 0.8989173769950867], "decide_seconds": 0.16152645100009977, "probabilities_f64_minus_runtime_max_abs": 2.2200479565803732e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.02106277751937444, "darwin": 0.006423784204603875, "sagan": 0.01359915126786264, "ibn_sina": 0.012775220344206605, "einstein": 0.8956120900503995, "socrates": 0.050526976613552946}, "confidence": 0.7335262106710523, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 165.31119100181968, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "51144324b9565a106912a27b203efc22e0622c9d2a6c6bf22976aed0d6f4c980", "compiled_ids_sha256": "b70e47b4e50f336958b0e276a601b674909a0444cca8245c459f864bb91e16c3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.875, 14.6875, 15.4375, 15.375, 19.625, 16.75], "runtime_probabilities": [0.021062780171632767, 0.0064237844198942184, 0.013599151745438576, 0.012775220908224583, 0.8956121802330017, 0.05052698403596878], "decide_seconds": 0.16151239100145176, "probabilities_f64_minus_runtime_max_abs": 9.018260216375751e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.02030965581355217, "einstein": 0.8635885402411037, "darwin": 0.0035292889808422184, "sagan": 0.007018829237184819, "hypatia": 0.04299554169468765, "socrates": 0.06255814403262944}, "confidence": 0.6823172822426213, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 165.51388999869232, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "9861ffca80b3747e09e69887723701d4efc811d7e5b8ba2e3cca712dcc6e7fd9", "compiled_ids_sha256": "a3c5e21cefdb935f6a684b8afb1807ec59a612ca14af69e48bb301e7359ed68f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 19.5, 14.0, 14.6875, 16.5, 16.875], "runtime_probabilities": [0.020309654995799065, 0.8635884523391724, 0.00352928857319057, 0.007018829230219126, 0.04299553483724594, 0.0625581368803978], "decide_seconds": 0.16158536100192578, "probabilities_f64_minus_runtime_max_abs": 8.790193128938029e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.8447506053814444, "sagan": 0.008815763644651052, "socrates": 0.037115751276977756, "ibn_sina": 0.028905776158793767, "darwin": 0.03275453553903311, "hypatia": 0.04765756799909993}, "confidence": 0.6283277774780722, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 165.44202099976246, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "a4cdd1d51e4bd6a3833600e0339a5be23818ee80cc31a34409d53b098778e816", "compiled_ids_sha256": "529c224f0eda1a82a9fdda9e8ffc63262bb8440329bcbfe19872bd59d1052dce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 15.0625, 16.5, 16.25, 16.375, 16.75], "runtime_probabilities": [0.8447505831718445, 0.008815763518214226, 0.03711574897170067, 0.028905773535370827, 0.03275453299283981, 0.047657568007707596], "decide_seconds": 0.16157766000105767, "probabilities_f64_minus_runtime_max_abs": 2.2209599936928726e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015455362247486998, "ibn_sina": 0.011307384571293383, "hypatia": 0.022487419379907853, "einstein": 0.8982568160992617, "sagan": 0.04472157351503373, "darwin": 0.007771444187016396}, "confidence": 0.7357022920886073, "choice": "einstein"}}, "usage": {"input_tokens": 316, "output_tokens": 1}, "latency_ms": 165.18621199793415, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 316, "compiled_prompt_sha256": "d6853c19ae77d785ff4d0e26d079aa1237f28072b8649be382018b78c39bc252", "compiled_ids_sha256": "c50647ddb58bcc8dc8ee0fc4a330c94e2ecf055f136663f079dabdd1c425c3eb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 15.125, 15.8125, 19.5, 16.5, 14.75], "runtime_probabilities": [0.015455360524356365, 0.011307383887469769, 0.022487418726086617, 0.898256778717041, 0.044721569865942, 0.0077714440412819386], "decide_seconds": 0.16137523199722636, "probabilities_f64_minus_runtime_max_abs": 3.7382220652304454e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:19Z"} {"id": "c-sagan-01-main-03", "uid": "c083", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "prompt_sha256_by_order": ["8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad"], "request_sha256": "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "compiled_prompt_sha256_by_order": ["24b38735ab28fcd79f74277c7c0d9977f3ca86b28f20b968104f2e28ecd092e0", "ce28dea2a0a54fcc0e167a92ae8994c68835cad18971eebd17e64ab5f569e2ef", "77511413acf0a4b81abae4ee583725750c165d808590e86e0620d3a48e4a0cd4", "43f42d70a8389070f921aa5b565f8a3579f4d4a0689dd5a24d78d8e1582ee5b0", "f717a4e8245349623daaaf57dd81e85e8bbbbfb3c80e328ea9dd74481098fbe9", "fafbafb95e330679ee2c3ab99b2174edbe94237eb1806c2fe432e0d7decbe8ce"], "compiled_tokens_by_order": [399, 399, 399, 399, 399, 399], "choice": "sagan", "probs": {"darwin": 0.008730936602459168, "einstein": 0.01053151983566309, "hypatia": 0.013522739145342515, "ibn_sina": 0.0038743296318296976, "sagan": 0.9480172038409344, "socrates": 0.015323270943771087}, "confidence": 0.8416697744185477, "seconds": 0.17731629633332582, "seconds_all_orders": 1.063897777999955, "seconds_by_order": [0.17747362799855182, 0.1774411689984845, 0.17723426000156906, 0.1771491209983651, 0.17729998000140768, 0.17729962000157684], "argmax_stable": true, "spread": {"darwin": 0.03602808833475188, "einstein": 0.11686887861902306, "hypatia": 0.02926601941821508, "ibn_sina": 0.01011681622325612, "sagan": 0.23426533514600456, "socrates": 0.07136577278704628}, "spread_2dp": {"darwin": 0.04, "einstein": 0.11, "hypatia": 0.03, "ibn_sina": 0.01, "sagan": 0.24, "socrates": 0.08}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.008730936602459168, "einstein": 0.01053151983566309, "hypatia": 0.013522739145342515, "ibn_sina": 0.0038743296318296976, "sagan": 0.9480172038409344, "socrates": 0.015323270943771087}, {"ibn_sina": 0.013991145855085816, "hypatia": 0.010896315348007075, "sagan": 0.8656015429565246, "socrates": 0.015854045282322832, "darwin": 0.013143465179996235, "einstein": 0.08051348537806333}, {"hypatia": 0.011522896324743367, "darwin": 0.004239036660623056, "sagan": 0.9153770352149548, "ibn_sina": 0.008974040680960747, "einstein": 0.024393971710907833, "socrates": 0.035493019407810175}, {"ibn_sina": 0.011093649185334519, "einstein": 0.007162598249219217, "darwin": 0.011093649185334519, "sagan": 0.9381152404807789, "hypatia": 0.01829033538154617, "socrates": 0.01424452751778678}, {"einstein": 0.12403147686824229, "sagan": 0.7137518686949298, "socrates": 0.08524550428415338, "ibn_sina": 0.012280779269123748, "darwin": 0.04026712499537494, "hypatia": 0.024423245888175828}, {"socrates": 0.013879731497107103, "ibn_sina": 0.0045060890768072225, "hypatia": 0.040162334766222156, "einstein": 0.017821928019086917, "sagan": 0.9140905259903176, "darwin": 0.00953939065045908}], "prompt_tokens": 399, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.008730936602459168, "einstein": 0.01053151983566309, "hypatia": 0.013522739145342515, "ibn_sina": 0.0038743296318296976, "sagan": 0.9480172038409344, "socrates": 0.015323270943771087}, "confidence": 0.8416697744185477, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.44579799909843, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "24b38735ab28fcd79f74277c7c0d9977f3ca86b28f20b968104f2e28ecd092e0", "compiled_ids_sha256": "c413e920def965f2042bf7df2f36f935761ed11a9ea1a07f8d6b709f90857e2d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.875, 15.125, 13.875, 19.375, 15.25], "runtime_probabilities": [0.008730936795473099, 0.01053152047097683, 0.013522738590836525, 0.0038743296172469854, 0.9480171799659729, 0.015323271043598652], "decide_seconds": 0.1727121250005439, "probabilities_f64_minus_runtime_max_abs": 2.387496145228596e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.013991145855085816, "hypatia": 0.010896315348007075, "sagan": 0.8656015429565246, "socrates": 0.015854045282322832, "darwin": 0.013143465179996235, "einstein": 0.08051348537806333}, "confidence": 0.6877988236627741, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.41539900089265, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "ce28dea2a0a54fcc0e167a92ae8994c68835cad18971eebd17e64ab5f569e2ef", "compiled_ids_sha256": "bf0094bac882a568df3304d91bbb1ca238e146cfd92ef576cc50e57d22116f4e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 15.25, 19.625, 15.625, 15.4375, 17.25], "runtime_probabilities": [0.013991144485771656, 0.01089631486684084, 0.8656014800071716, 0.015854043886065483, 0.013143463991582394, 0.08051347732543945], "decide_seconds": 0.17272150499775307, "probabilities_f64_minus_runtime_max_abs": 6.294935295247939e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.011522896324743367, "darwin": 0.004239036660623056, "sagan": 0.9153770352149548, "ibn_sina": 0.008974040680960747, "einstein": 0.024393971710907833, "socrates": 0.035493019407810175}, "confidence": 0.7729035970661138, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.208380002412, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "77511413acf0a4b81abae4ee583725750c165d808590e86e0620d3a48e4a0cd4", "compiled_ids_sha256": "1840e4a628ef8a1d24dc10b679fe046b93574447ba930937e0a19cfe31b59fa5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 14.375, 19.75, 15.125, 16.125, 16.5], "runtime_probabilities": [0.011522895656526089, 0.004239036701619625, 0.9153769612312317, 0.008974040858447552, 0.024393970146775246, 0.03549301624298096], "decide_seconds": 0.17261617600161117, "probabilities_f64_minus_runtime_max_abs": 7.398372314781909e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.011093649185334519, "einstein": 0.007162598249219217, "darwin": 0.011093649185334519, "sagan": 0.9381152404807789, "hypatia": 0.01829033538154617, "socrates": 0.01424452751778678}, "confidence": 0.8164243342668369, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.1243909970508, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "43f42d70a8389070f921aa5b565f8a3579f4d4a0689dd5a24d78d8e1582ee5b0", "compiled_ids_sha256": "0765e26d6f121f66398db5c61a9d90108bf60fa208ecde1dc2c3d3b33800173e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 14.75, 15.1875, 19.625, 15.6875, 15.4375], "runtime_probabilities": [0.011093649081885815, 0.0071625979617238045, 0.011093649081885815, 0.9381152987480164, 0.018290337175130844, 0.01424452755600214], "decide_seconds": 0.1724998770005186, "probabilities_f64_minus_runtime_max_abs": 5.826723747937024e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.12403147686824229, "sagan": 0.7137518686949298, "socrates": 0.08524550428415338, "ibn_sina": 0.012280779269123748, "darwin": 0.04026712499537494, "hypatia": 0.024423245888175828}, "confidence": 0.4510932952172829, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.27336000098148, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "f717a4e8245349623daaaf57dd81e85e8bbbbfb3c80e328ea9dd74481098fbe9", "compiled_ids_sha256": "aea9ce8aef233bdbabac501091443650f0703f298c762ae00bddccb5f19e513b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 19.125, 17.0, 15.0625, 16.25, 15.75], "runtime_probabilities": [0.12403147667646408, 0.7137519121170044, 0.0852455124258995, 0.012280779890716076, 0.040267132222652435, 0.024423247203230858], "decide_seconds": 0.17258673700052896, "probabilities_f64_minus_runtime_max_abs": 4.3422074602617045e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.013879731497107103, "ibn_sina": 0.0045060890768072225, "hypatia": 0.040162334766222156, "einstein": 0.017821928019086917, "sagan": 0.9140905259903176, "darwin": 0.00953939065045908}, "confidence": 0.7705661629440048, "choice": "sagan"}}, "usage": {"input_tokens": 399, "output_tokens": 1}, "latency_ms": 177.26920999848517, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 399, "compiled_prompt_sha256": "fafbafb95e330679ee2c3ab99b2174edbe94237eb1806c2fe432e0d7decbe8ce", "compiled_ids_sha256": "2982ffbcaea14e2fe889c678fa398fb5e9810eba97ed7f717e43a428934f4889", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.3125, 14.1875, 16.375, 15.5625, 19.5, 14.9375], "runtime_probabilities": [0.013879731297492981, 0.004506089258939028, 0.0401623360812664, 0.01782192848622799, 0.9140905141830444, 0.009539390914142132], "decide_seconds": 0.17258923599729314, "probabilities_f64_minus_runtime_max_abs": 1.1807273159192277e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:21Z"} {"id": "c-sagan-01-main-03", "uid": "c084", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "prompt_sha256_by_order": ["6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726"], "request_sha256": "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "compiled_prompt_sha256_by_order": ["18fc51ab32b6e239ab8298165343e7ff14bc85e0d166e6dc0b5e9964cae0940b", "55e53c7efb5da7ac634f55373a8eede9f8afa98982adeda8d85d7fdf5b2f8305", "38f455cfd986d076d286f2f4dac7050010666dd4fe22f1ff3e497ecfd95f5e8e", "a1ffcaa9ca0dbb7f9a7d2ff5069d542627fbde9107efb777269f8e5f71022423", "6ba87c856e7f4a01dce52756c360eb546ab93e9bd4f614569d416638894f9a4f", "380b1f584093c36086b4f2f4ddc88cc74acf880709274ca41d30bdb74ddc8d43"], "compiled_tokens_by_order": [400, 400, 400, 400, 400, 400], "choice": "sagan", "probs": {"darwin": 0.007997529104976371, "einstein": 0.19376239339438514, "hypatia": 0.009062387733660896, "ibn_sina": 0.013185696348417902, "sagan": 0.7663451318915124, "socrates": 0.009646861527047202}, "confidence": 0.6065203323218453, "seconds": 0.1771150911681616, "seconds_all_orders": 1.0626905470089696, "seconds_by_order": [0.17747762900035013, 0.17758335700273165, 0.17687301300247782, 0.17683050300183822, 0.1770365420015878, 0.17688950299998396], "argmax_stable": true, "spread": {"darwin": 0.013222485118464803, "einstein": 0.3435205041908534, "hypatia": 0.007708121522571869, "ibn_sina": 0.029080274138747864, "sagan": 0.42239716169924224, "socrates": 0.046105971446432754}, "spread_2dp": {"darwin": 0.01, "einstein": 0.33999999999999997, "hypatia": 0.01, "ibn_sina": 0.03, "sagan": 0.41999999999999993, "socrates": 0.05}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.007997529104976371, "einstein": 0.19376239339438514, "hypatia": 0.009062387733660896, "ibn_sina": 0.013185696348417902, "sagan": 0.7663451318915124, "socrates": 0.009646861527047202}, {"ibn_sina": 0.004985234807239437, "hypatia": 0.0022121845358350706, "sagan": 0.9500175946056969, "socrates": 0.003023697756457318, "darwin": 0.007253473844389203, "einstein": 0.032507814450382116}, {"hypatia": 0.004631983197954933, "darwin": 0.017209905390972446, "sagan": 0.9396289965925316, "ibn_sina": 0.015187688201307116, "einstein": 0.019501377671206865, "socrates": 0.003840048946027008}, {"ibn_sina": 0.00962460045988306, "einstein": 0.04591640077714476, "darwin": 0.006214109104380709, "sagan": 0.922255563189224, "hypatia": 0.007495646374906329, "socrates": 0.008493680094461073}, {"einstein": 0.36302188186206025, "sagan": 0.5281937214278405, "socrates": 0.049129669202890074, "ibn_sina": 0.03376629491369335, "darwin": 0.0192394548966216, "hypatia": 0.0066489776968942154}, {"socrates": 0.004988243149303448, "ibn_sina": 0.00468602077494549, "hypatia": 0.00992030605840694, "einstein": 0.023797577112104534, "sagan": 0.9505908831270827, "darwin": 0.006016969778156796}], "prompt_tokens": 400, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.007997529104976371, "einstein": 0.19376239339438514, "hypatia": 0.009062387733660896, "ibn_sina": 0.013185696348417902, "sagan": 0.7663451318915124, "socrates": 0.009646861527047202}, "confidence": 0.6065203323218453, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 177.44668900195393, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "18fc51ab32b6e239ab8298165343e7ff14bc85e0d166e6dc0b5e9964cae0940b", "compiled_ids_sha256": "73adbcdaa744c056b6a9cf97351cb9d64087fb324180be15d21ff25d3141cbfd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 18.0, 14.9375, 15.3125, 19.375, 15.0], "runtime_probabilities": [0.007997529581189156, 0.19376240670681, 0.009062387980520725, 0.013185696676373482, 0.7663451433181763, 0.00964686181396246], "decide_seconds": 0.17276444399976754, "probabilities_f64_minus_runtime_max_abs": 1.3312424856914973e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004985234807239437, "hypatia": 0.0022121845358350706, "sagan": 0.9500175946056969, "socrates": 0.003023697756457318, "darwin": 0.007253473844389203, "einstein": 0.032507814450382116}, "confidence": 0.8586197251251635, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 177.55459800173412, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "55e53c7efb5da7ac634f55373a8eede9f8afa98982adeda8d85d7fdf5b2f8305", "compiled_ids_sha256": "082cb715241a64ca4a2624f19ffc2130c7fb6a9bfe2ee77e84249a9596f63472", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 13.8125, 19.875, 14.125, 15.0, 16.5], "runtime_probabilities": [0.004985234700143337, 0.0022121844813227654, 0.9500176310539246, 0.0030236979946494102, 0.007253473624587059, 0.03250781446695328], "decide_seconds": 0.1727139750000788, "probabilities_f64_minus_runtime_max_abs": 3.6448227658247845e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.004631983197954933, "darwin": 0.017209905390972446, "sagan": 0.9396289965925316, "ibn_sina": 0.015187688201307116, "einstein": 0.019501377671206865, "socrates": 0.003840048946027008}, "confidence": 0.8241647244924213, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 176.84545299925958, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "38f455cfd986d076d286f2f4dac7050010666dd4fe22f1ff3e497ecfd95f5e8e", "compiled_ids_sha256": "85e9a30fa7702e0d8709a5785cef96e67c06280ab7efd776837cf4537b2873c3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.4375, 15.75, 19.75, 15.625, 15.875, 14.25], "runtime_probabilities": [0.004631983581930399, 0.01720990613102913, 0.939629077911377, 0.01518768910318613, 0.019501378759741783, 0.0038400492630898952], "decide_seconds": 0.17224587900273036, "probabilities_f64_minus_runtime_max_abs": 8.131884532414091e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.00962460045988306, "einstein": 0.04591640077714476, "darwin": 0.006214109104380709, "sagan": 0.922255563189224, "hypatia": 0.007495646374906329, "socrates": 0.008493680094461073}, "confidence": 0.7937492014125243, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 176.8028240003332, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "a1ffcaa9ca0dbb7f9a7d2ff5069d542627fbde9107efb777269f8e5f71022423", "compiled_ids_sha256": "74d152d484047691c38ac5fdebfb1e0524429e0edb578d3126afff2e3aefc7c7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 16.625, 14.625, 19.625, 14.8125, 14.9375], "runtime_probabilities": [0.009624600410461426, 0.045916393399238586, 0.006214108318090439, 0.9222554564476013, 0.0074956463649868965, 0.008493678644299507], "decide_seconds": 0.1721986789998482, "probabilities_f64_minus_runtime_max_abs": 1.067416226518958e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.36302188186206025, "sagan": 0.5281937214278405, "socrates": 0.049129669202890074, "ibn_sina": 0.03376629491369335, "darwin": 0.0192394548966216, "hypatia": 0.0066489776968942154}, "confidence": 0.3990345036529763, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 177.0092820006539, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "6ba87c856e7f4a01dce52756c360eb546ab93e9bd4f614569d416638894f9a4f", "compiled_ids_sha256": "7cc99a50550515f05096a6fe74c9707656dbd6595edb8c239ebc990f28b21f5f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.5, 18.875, 16.5, 16.125, 15.5625, 14.5], "runtime_probabilities": [0.3630219101905823, 0.5281937122344971, 0.049129668623209, 0.03376629576086998, 0.019239455461502075, 0.006648977752774954], "decide_seconds": 0.17239032799989218, "probabilities_f64_minus_runtime_max_abs": 2.8328522028164826e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.004988243149303448, "ibn_sina": 0.00468602077494549, "hypatia": 0.00992030605840694, "einstein": 0.023797577112104534, "sagan": 0.9505908831270827, "darwin": 0.006016969778156796}, "confidence": 0.8519722410014132, "choice": "sagan"}}, "usage": {"input_tokens": 400, "output_tokens": 1}, "latency_ms": 176.8622330018843, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 400, "compiled_prompt_sha256": "380b1f584093c36086b4f2f4ddc88cc74acf880709274ca41d30bdb74ddc8d43", "compiled_ids_sha256": "f3ec31b87dd977f9e983b20c34c4b2e7dced1ed7adf3d950f5d6ee2845f9b206", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 14.3125, 15.0625, 15.9375, 19.625, 14.5625], "runtime_probabilities": [0.0049882433377206326, 0.004686020780354738, 0.009920306503772736, 0.023797577247023582, 0.9505908489227295, 0.006016969680786133], "decide_seconds": 0.1722943690001557, "probabilities_f64_minus_runtime_max_abs": 3.420435323864268e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:22Z"} {"id": "c-sagan-01-starter-01", "uid": "c085", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "prompt_sha256_by_order": ["895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655"], "request_sha256": "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "compiled_prompt_sha256_by_order": ["65b093d4dc7322a71d795f378d96f989360cca3b9cc8cc2dafc040037e482792", "29a9bb31b220de972a1c5ad69356b5d1101174b43840ce57d6315d8082c32b05", "ce7c4dc2c035eee38fbc626a23e11fb9ededff2b103befafff23d764bfad4c17", "f37681dc34ec37cec21c16f17078a46c8da2e6e2793ff9b7f85f423475ce1c3e", "152f3406252b1770dd3a9f9aa80f49ee5566c556d5802738b80d336bff5724bf", "38f69caf3a1bd95791d66f62f117a483a70022cbbb21d3824557a66bd1a6fc15"], "compiled_tokens_by_order": [284, 284, 284, 284, 284, 284], "choice": "sagan", "probs": {"darwin": 0.0014026335585843907, "einstein": 0.0014930956510053226, "hypatia": 0.0014930956510053226, "ibn_sina": 0.0007991965110726008, "sagan": 0.9931200796011149, "socrates": 0.0016918990272174873}, "confidence": 0.9709788871282936, "seconds": 0.1650523919997795, "seconds_all_orders": 0.990314351998677, "seconds_by_order": [0.16534868099915911, 0.16510882299917284, 0.16504196399910143, 0.16506990400012, 0.16495079399828683, 0.1647941860028368], "argmax_stable": true, "spread": {"darwin": 0.0008845922980483937, "einstein": 0.000828263772942032, "hypatia": 0.0010066942583922712, "ibn_sina": 0.0006120328566859873, "sagan": 0.003928618744843959, "socrates": 0.0011404469758860953}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.010000000000000009, "socrates": 0.0}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0014026335585843907, "einstein": 0.0014930956510053226, "hypatia": 0.0014930956510053226, "ibn_sina": 0.0007991965110726008, "sagan": 0.9931200796011149, "socrates": 0.0016918990272174873}, {"ibn_sina": 0.0006245517508274712, "hypatia": 0.00048640139261305147, "sagan": 0.9965296713441907, "socrates": 0.0006648318780632906, "darwin": 0.001029711756242258, "einstein": 0.0006648318780632906}, {"hypatia": 0.0006651781454211764, "darwin": 0.000518041260535997, "sagan": 0.9970486983459589, "ibn_sina": 0.0005514520513313921, "einstein": 0.0006651781454211764, "socrates": 0.0005514520513313921}, {"ibn_sina": 0.0011634849080173794, "einstein": 0.0014034306907953504, "darwin": 0.0013184011236864921, "sagan": 0.9936844807591763, "hypatia": 0.0014034306907953504, "socrates": 0.00102677182752926}, {"einstein": 0.0014048581138232474, "sagan": 0.9946951527642867, "socrates": 0.0011646682832714673, "ibn_sina": 0.0006636079843169314, "darwin": 0.0011646682832714673, "hypatia": 0.0009070445710302476}, {"socrates": 0.0010288820093255157, "ibn_sina": 0.00070713957415034, "hypatia": 0.0010288820093255157, "einstein": 0.0008012941145507921, "sagan": 0.9957266627184975, "darwin": 0.00070713957415034}], "prompt_tokens": 284, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0014026335585843907, "einstein": 0.0014930956510053226, "hypatia": 0.0014930956510053226, "ibn_sina": 0.0007991965110726008, "sagan": 0.9931200796011149, "socrates": 0.0016918990272174873}, "confidence": 0.9709788871282936, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 165.32302199993865, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "65b093d4dc7322a71d795f378d96f989360cca3b9cc8cc2dafc040037e482792", "compiled_ids_sha256": "c711190daf3352b61206e8801960f56744ba55154d91e3a74652ef38d6ba1604", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.9375, 13.0, 13.0, 12.375, 19.5, 13.125], "runtime_probabilities": [0.001402633497491479, 0.0014930955367162824, 0.0014930955367162824, 0.0007991964812390506, 0.9931200742721558, 0.0016918991459533572], "decide_seconds": 0.1617383200027689, "probabilities_f64_minus_runtime_max_abs": 5.3289591628313815e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0006245517508274712, "hypatia": 0.00048640139261305147, "sagan": 0.9965296713441907, "socrates": 0.0006648318780632906, "darwin": 0.001029711756242258, "einstein": 0.0006648318780632906}, "confidence": 0.9840415641569835, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 165.08624299967778, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "29a9bb31b220de972a1c5ad69356b5d1101174b43840ce57d6315d8082c32b05", "compiled_ids_sha256": "b34c05271926877c4d55edccdd464343b298f3cb911940c7a6a56c5f09eb4aee", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.5, 12.25, 19.875, 12.5625, 13.0, 12.5625], "runtime_probabilities": [0.0006245517288334668, 0.00048640146269463, 0.996529757976532, 0.0006648319540545344, 0.0010297117987647653, 0.0006648319540545344], "decide_seconds": 0.16166376999899512, "probabilities_f64_minus_runtime_max_abs": 8.663234130068531e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0006651781454211764, "darwin": 0.000518041260535997, "sagan": 0.9970486983459589, "ibn_sina": 0.0005514520513313921, "einstein": 0.0006651781454211764, "socrates": 0.0005514520513313921}, "confidence": 0.9861179115058085, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 165.01947399956407, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "ce7c4dc2c035eee38fbc626a23e11fb9ededff2b103befafff23d764bfad4c17", "compiled_ids_sha256": "429fe0a0e84aa69822311049d513022e69e24731fa215f45eff10b1d3c6ce3b3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.5625, 12.3125, 19.875, 12.375, 12.5625, 12.375], "runtime_probabilities": [0.0006651781150139868, 0.0005180412554182112, 0.9970486760139465, 0.0005514520453289151, 0.0006651781150139868, 0.0005514520453289151], "decide_seconds": 0.16150073199969484, "probabilities_f64_minus_runtime_max_abs": 2.2332012350467778e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0011634849080173794, "einstein": 0.0014034306907953504, "darwin": 0.0013184011236864921, "sagan": 0.9936844807591763, "hypatia": 0.0014034306907953504, "socrates": 0.00102677182752926}, "confidence": 0.9729860162402117, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 165.0480440002866, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "f37681dc34ec37cec21c16f17078a46c8da2e6e2793ff9b7f85f423475ce1c3e", "compiled_ids_sha256": "b7269e259ce88d0455d1f3329a83764530595bc0df657ec82d7fbcf0ead6697e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.875, 13.0625, 13.0, 19.625, 13.0625, 12.75], "runtime_probabilities": [0.0011634847614914179, 0.001403430593200028, 0.0013184009585529566, 0.9936843514442444, 0.001403430593200028, 0.001026771729812026], "decide_seconds": 0.16165698999975575, "probabilities_f64_minus_runtime_max_abs": 1.2931493187373633e-07}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0014048581138232474, "sagan": 0.9946951527642867, "socrates": 0.0011646682832714673, "ibn_sina": 0.0006636079843169314, "darwin": 0.0011646682832714673, "hypatia": 0.0009070445710302476}, "confidence": 0.9768588824619903, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 164.92839499915135, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "152f3406252b1770dd3a9f9aa80f49ee5566c556d5802738b80d336bff5724bf", "compiled_ids_sha256": "a29acbd8841f67c53988645b61dd30d45fa4aa082bdad89c04b9d96ca097a6ac", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.1875, 19.75, 13.0, 12.4375, 13.0, 12.75], "runtime_probabilities": [0.0014048581942915916, 0.994695246219635, 0.0011646683560684323, 0.000663608021568507, 0.0011646683560684323, 0.0009070446249097586], "decide_seconds": 0.16146070200193208, "probabilities_f64_minus_runtime_max_abs": 9.345534834892533e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.0010288820093255157, "ibn_sina": 0.00070713957415034, "hypatia": 0.0010288820093255157, "einstein": 0.0008012941145507921, "sagan": 0.9957266627184975, "darwin": 0.00070713957415034}, "confidence": 0.9808052550057365, "choice": "sagan"}}, "usage": {"input_tokens": 284, "output_tokens": 1}, "latency_ms": 164.77244600173435, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 284, "compiled_prompt_sha256": "38f69caf3a1bd95791d66f62f117a483a70022cbbb21d3824557a66bd1a6fc15", "compiled_ids_sha256": "26d971bdc909a84a904f4e9c530caa375291f3369e7a3de38e9e1d25f10ee900", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [12.75, 12.375, 12.75, 12.5, 19.625, 12.375], "runtime_probabilities": [0.0010288819903507829, 0.0007071396103128791, 0.0010288819903507829, 0.0008012941689230502, 0.9957267045974731, 0.0007071396103128791], "decide_seconds": 0.16141297099966323, "probabilities_f64_minus_runtime_max_abs": 4.187897562069054e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:23Z"} {"id": "c-sagan-01-starter-02", "uid": "c086", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "prompt_sha256_by_order": ["30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263"], "request_sha256": "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "compiled_prompt_sha256_by_order": ["e2b81005504d2eebe6b4f6eae69c57ba10486928d37eddc53c160a4b2854c594", "9f7e16f19df496ad1969d45d35fdba9da41a87f38a37323bc74379e8bafeaa71", "6403a7bcb3e90824947545b04e0ce62f413fe159bd666c42fbb3771e5d9f598f", "b6b13f14b5cd2091588d3644643557e4d50a01260a02fce23b08a21c3144f3c8", "b1eddab09bd632438871f1817cb5d6b314a85f8d31b1c6c3bfa99dbc7399640c", "230ef714e3cb334d5bc4c4ea4dff194999629797df91836c38b3f68ebc471b30"], "compiled_tokens_by_order": [404, 404, 404, 404, 404, 404], "choice": "sagan", "probs": {"darwin": 0.030852273339752122, "einstein": 0.027227035660024542, "hypatia": 0.02402777463653198, "ibn_sina": 0.004175401150928748, "sagan": 0.9016355913771451, "socrates": 0.012081923835617528}, "confidence": 0.7406947431638271, "seconds": 0.1776351524998366, "seconds_all_orders": 1.0658109149990196, "seconds_by_order": [0.17780755599960685, 0.17787391599995317, 0.1776158079992456, 0.17749641800037352, 0.17745906900017872, 0.17755814799966174], "argmax_stable": false, "spread": {"darwin": 0.03016336371480853, "einstein": 0.5545337780352142, "hypatia": 0.01737373219939182, "ibn_sina": 0.010044195721150357, "sagan": 0.5942415394191768, "socrates": 0.020317170875681766}, "spread_2dp": {"darwin": 0.030000000000000002, "einstein": 0.5499999999999999, "hypatia": 0.019999999999999997, "ibn_sina": 0.01, "sagan": 0.5900000000000001, "socrates": 0.019999999999999997}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "einstein", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.030852273339752122, "einstein": 0.027227035660024542, "hypatia": 0.02402777463653198, "ibn_sina": 0.004175401150928748, "sagan": 0.9016355913771451, "socrates": 0.012081923835617528}, {"ibn_sina": 0.014219596872079106, "hypatia": 0.016112914198830323, "sagan": 0.68513854590844, "socrates": 0.02344415182377783, "darwin": 0.03865287178538571, "einstein": 0.22243191941148704}, {"hypatia": 0.01322173277171391, "darwin": 0.016977040931533918, "sagan": 0.8707560853261748, "ibn_sina": 0.009673226337541897, "einstein": 0.06307735925712403, "socrates": 0.026294555375911584}, {"ibn_sina": 0.00876574747638971, "einstein": 0.019753916873970604, "darwin": 0.0270004030701363, "sagan": 0.894130550734427, "hypatia": 0.03059546497110573, "socrates": 0.019753916873970604}, {"einstein": 0.5742876949091849, "sagan": 0.3073940519579683, "socrates": 0.032399094711299294, "ibn_sina": 0.011918960856853417, "darwin": 0.04714040464634245, "hypatia": 0.026859792918351725}, {"socrates": 0.012597202321262014, "ibn_sina": 0.005950497034675645, "hypatia": 0.023534672311653564, "einstein": 0.05645674396003726, "sagan": 0.8831320631467269, "darwin": 0.018328821225644706}], "prompt_tokens": 404, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.030852273339752122, "einstein": 0.027227035660024542, "hypatia": 0.02402777463653198, "ibn_sina": 0.004175401150928748, "sagan": 0.9016355913771451, "socrates": 0.012081923835617528}, "confidence": 0.7406947431638271, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.77861599824973, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "e2b81005504d2eebe6b4f6eae69c57ba10486928d37eddc53c160a4b2854c594", "compiled_ids_sha256": "6f3ebc0833ff0d4eca3500ecdd2bf006ccd0dd84cbb8ec6be2d67191859098b9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 16.0, 15.875, 14.125, 19.5, 15.1875], "runtime_probabilities": [0.03085227496922016, 0.027227038517594337, 0.02402777597308159, 0.004175401758402586, 0.9016356468200684, 0.012081924825906754], "decide_seconds": 0.17291491400101222, "probabilities_f64_minus_runtime_max_abs": 5.5442923252435605e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.014219596872079106, "hypatia": 0.016112914198830323, "sagan": 0.68513854590844, "socrates": 0.02344415182377783, "darwin": 0.03865287178538571, "einstein": 0.22243191941148704}, "confidence": 0.47864347047432965, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.8476059989771, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "9f7e16f19df496ad1969d45d35fdba9da41a87f38a37323bc74379e8bafeaa71", "compiled_ids_sha256": "ebc29f3be1769f4245904f8ffdbe8182607653d61981ed695aa218af9787ded6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 15.625, 19.375, 16.0, 16.5, 18.25], "runtime_probabilities": [0.014219596050679684, 0.0161129143089056, 0.6851385235786438, 0.023444151505827904, 0.038652874529361725, 0.22243191301822662], "decide_seconds": 0.1731320320031955, "probabilities_f64_minus_runtime_max_abs": 2.2329796234288324e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.01322173277171391, "darwin": 0.016977040931533918, "sagan": 0.8707560853261748, "ibn_sina": 0.009673226337541897, "einstein": 0.06307735925712403, "socrates": 0.026294555375911584}, "confidence": 0.6864841370513973, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.59012800161145, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "6403a7bcb3e90824947545b04e0ce62f413fe159bd666c42fbb3771e5d9f598f", "compiled_ids_sha256": "6e2c14c80d83f33fd3c4523a18516f272ecdf355dada430a4e086c4467b2d1a5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.8125, 19.75, 15.25, 17.125, 16.25], "runtime_probabilities": [0.013221732340753078, 0.016977041959762573, 0.8707560896873474, 0.00967322662472725, 0.06307736039161682, 0.026294555515050888], "decide_seconds": 0.17276848400069866, "probabilities_f64_minus_runtime_max_abs": 4.361172645239719e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.00876574747638971, "einstein": 0.019753916873970604, "darwin": 0.0270004030701363, "sagan": 0.894130550734427, "hypatia": 0.03059546497110573, "socrates": 0.019753916873970604}, "confidence": 0.720481573318136, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.47133800003212, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "b6b13f14b5cd2091588d3644643557e4d50a01260a02fce23b08a21c3144f3c8", "compiled_ids_sha256": "e9fea21702e5747b46432936efcbde9d65200b9e1bc6e57d7471bd0a9431b79a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 15.8125, 16.125, 19.625, 16.25, 15.8125], "runtime_probabilities": [0.00876574870198965, 0.019753916189074516, 0.02700040489435196, 0.8941305875778198, 0.03059546649456024, 0.019753916189074516], "decide_seconds": 0.17278812500080676, "probabilities_f64_minus_runtime_max_abs": 3.684339278109405e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.5742876949091849, "sagan": 0.3073940519579683, "socrates": 0.032399094711299294, "ibn_sina": 0.011918960856853417, "darwin": 0.04714040464634245, "hypatia": 0.026859792918351725}, "confidence": 0.39378661212334143, "choice": "einstein"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.43287900157156, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "b1eddab09bd632438871f1817cb5d6b314a85f8d31b1c6c3bfa99dbc7399640c", "compiled_ids_sha256": "f49f16cb89f90e7089d843c9b1c735f8b93a6df46048d0a8d32b71d189ed425b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.0, 18.375, 16.125, 15.125, 16.5, 15.9375], "runtime_probabilities": [0.5742876529693604, 0.3073940575122833, 0.032399095594882965, 0.011918960139155388, 0.04714040458202362, 0.026859793812036514], "decide_seconds": 0.1727255250007147, "probabilities_f64_minus_runtime_max_abs": 4.193982450217959e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.012597202321262014, "ibn_sina": 0.005950497034675645, "hypatia": 0.023534672311653564, "einstein": 0.05645674396003726, "sagan": 0.8831320631467269, "darwin": 0.018328821225644706}, "confidence": 0.7102487544237721, "choice": "sagan"}}, "usage": {"input_tokens": 404, "output_tokens": 1}, "latency_ms": 177.53322900171042, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 404, "compiled_prompt_sha256": "230ef714e3cb334d5bc4c4ea4dff194999629797df91836c38b3f68ebc471b30", "compiled_ids_sha256": "843cb8d51c5a7617174e7c789eddfe37c09f43daae11d33fb7de9a493746acf2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 14.5, 15.875, 16.75, 19.5, 15.625], "runtime_probabilities": [0.012597201392054558, 0.00595049699768424, 0.02353467047214508, 0.05645674094557762, 0.8831320405006409, 0.018328819423913956], "decide_seconds": 0.17290165400117985, "probabilities_f64_minus_runtime_max_abs": 2.2646086006439248e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:24Z"} {"id": "c-sagan-01-starter-02", "uid": "c087", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "prompt_sha256_by_order": ["dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a"], "request_sha256": "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "compiled_prompt_sha256_by_order": ["260bb13ff919a69d62276bf87d96d93384b67dfa42212c818966dc0bd3ee540c", "dc8f222ffc40c32ac80db98d97b1f8417747db7cd32c674f41221704a9d6ba61", "a0c465baebf66bf5c8365e0f484b66df78bb7fd324ee6383fce1fed93160f6ba", "a941b2dd0565dce93f296a6e61caa6b3a5a993226d375855a21cf35282d92283", "6b7d821c7e72bc68abb3b4f45a0262758e0eb531ab19382b8c49de1d06fb8603", "78775807212758e40edf769f9eb7a399a8186be9fa4466e97223edc44ad77b5e"], "compiled_tokens_by_order": [293, 293, 293, 293, 293, 293], "choice": "sagan", "probs": {"darwin": 0.00468577472854613, "einstein": 0.006404694682770197, "hypatia": 0.01273725629685956, "ibn_sina": 0.001834975749038615, "sagan": 0.9505409709573996, "socrates": 0.023796327585385996}, "confidence": 0.8538935366695446, "seconds": 0.16502901716679239, "seconds_all_orders": 0.9901741030007543, "seconds_by_order": [0.16531632199985324, 0.1649864040009561, 0.16541921099997126, 0.16501522399994428, 0.16469903600227553, 0.16473790599775384], "argmax_stable": true, "spread": {"darwin": 0.009379077154758449, "einstein": 0.02302018461276397, "hypatia": 0.013188439519311113, "ibn_sina": 0.003070885205012438, "sagan": 0.029414046653229775, "socrates": 0.019369055177681012}, "spread_2dp": {"darwin": 0.01, "einstein": 0.019999999999999997, "hypatia": 0.02, "ibn_sina": 0.0, "sagan": 0.029999999999999916, "socrates": 0.019999999999999997}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.00468577472854613, "einstein": 0.006404694682770197, "hypatia": 0.01273725629685956, "ibn_sina": 0.001834975749038615, "sagan": 0.9505409709573996, "socrates": 0.023796327585385996}, {"ibn_sina": 0.004905860954051053, "hypatia": 0.004067101580652602, "sagan": 0.9348916155904036, "socrates": 0.02198654341836887, "darwin": 0.005917597882229652, "einstein": 0.028231280574294214}, {"hypatia": 0.0038980863311615563, "darwin": 0.002364308874056218, "sagan": 0.953830276504667, "ibn_sina": 0.0028519008828610695, "einstein": 0.008252248827826655, "socrates": 0.0288031785794275}, {"ibn_sina": 0.00489537161783606, "einstein": 0.005211095961530242, "darwin": 0.011743386028814666, "sagan": 0.9328927019293044, "hypatia": 0.01708652585047267, "socrates": 0.02817091861204189}, {"einstein": 0.008253540347776119, "sagan": 0.9539795559139783, "socrates": 0.023882410875382964, "ibn_sina": 0.0025171875869332, "darwin": 0.0053288861633548025, "hypatia": 0.006038419112574711}, {"socrates": 0.009434123401746491, "ibn_sina": 0.0025391598670406156, "hypatia": 0.012113654232011111, "einstein": 0.00886253875979496, "sagan": 0.9623067485825342, "darwin": 0.0047437751568727695}], "prompt_tokens": 293, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.00468577472854613, "einstein": 0.006404694682770197, "hypatia": 0.01273725629685956, "ibn_sina": 0.001834975749038615, "sagan": 0.9505409709573996, "socrates": 0.023796327585385996}, "confidence": 0.8538935366695446, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 165.28890200061142, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "260bb13ff919a69d62276bf87d96d93384b67dfa42212c818966dc0bd3ee540c", "compiled_ids_sha256": "56677b4e63c956b87e199aa8482bbf7c197f78ec3d9d566703cd84fe0d74f83f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0625, 14.375, 15.0625, 13.125, 19.375, 15.6875], "runtime_probabilities": [0.00468577491119504, 0.006404695101082325, 0.012737257406115532, 0.0018349757883697748, 0.9505410194396973, 0.023796329274773598], "decide_seconds": 0.16152411100119934, "probabilities_f64_minus_runtime_max_abs": 4.848229762099976e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004905860954051053, "hypatia": 0.004067101580652602, "sagan": 0.9348916155904036, "socrates": 0.02198654341836887, "darwin": 0.005917597882229652, "einstein": 0.028231280574294214}, "confidence": 0.8178258995504912, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 164.96166499928222, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "dc8f222ffc40c32ac80db98d97b1f8417747db7cd32c674f41221704a9d6ba61", "compiled_ids_sha256": "af59a0ac3a4a9fd765bbceaec5d8a816f181f880d3da6a5d1d767a3b5090c0ed", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 14.0625, 19.5, 15.75, 14.4375, 16.0], "runtime_probabilities": [0.0049058604054152966, 0.004067101515829563, 0.9348915219306946, 0.021986542269587517, 0.005917597096413374, 0.02823127992451191], "decide_seconds": 0.16145214200150804, "probabilities_f64_minus_runtime_max_abs": 9.365970898844722e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0038980863311615563, "darwin": 0.002364308874056218, "sagan": 0.953830276504667, "ibn_sina": 0.0028519008828610695, "einstein": 0.008252248827826655, "socrates": 0.0288031785794275}, "confidence": 0.8663431986215564, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 165.39485099929152, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "a0c465baebf66bf5c8365e0f484b66df78bb7fd324ee6383fce1fed93160f6ba", "compiled_ids_sha256": "4d05d2b9672ba78fc309e45b366de89998c5443bfdcf55ee59864853fec964b9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.125, 13.625, 19.625, 13.8125, 14.875, 16.125], "runtime_probabilities": [0.0038980862591415644, 0.0023643088061362505, 0.9538302421569824, 0.002851900877431035, 0.008252248167991638, 0.028803179040551186], "decide_seconds": 0.16169059900130378, "probabilities_f64_minus_runtime_max_abs": 3.4347684585434024e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.00489537161783606, "einstein": 0.005211095961530242, "darwin": 0.011743386028814666, "sagan": 0.9328927019293044, "hypatia": 0.01708652585047267, "socrates": 0.02817091861204189}, "confidence": 0.8099521796302991, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 164.99049400226795, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "a941b2dd0565dce93f296a6e61caa6b3a5a993226d375855a21cf35282d92283", "compiled_ids_sha256": "439ba1852928ce287bc969fdbf4ef486b2bd9cdc334691a62b1240ab5ee705dc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 14.3125, 15.125, 19.5, 15.5, 16.0], "runtime_probabilities": [0.004895371850579977, 0.005211096256971359, 0.011743387207388878, 0.9328927397727966, 0.01708652637898922, 0.02817092090845108], "decide_seconds": 0.16136960200310568, "probabilities_f64_minus_runtime_max_abs": 3.784349222879513e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.008253540347776119, "sagan": 0.9539795559139783, "socrates": 0.023882410875382964, "ibn_sina": 0.0025171875869332, "darwin": 0.0053288861633548025, "hypatia": 0.006038419112574711}, "confidence": 0.8618438224234286, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 164.6754069988674, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "6b7d821c7e72bc68abb3b4f45a0262758e0eb531ab19382b8c49de1d06fb8603", "compiled_ids_sha256": "b61a5716f9e55c7e3dd16078e1b806b15a10d5c0ddc7af3450078fb023c9391b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 19.375, 15.6875, 13.4375, 14.1875, 14.3125], "runtime_probabilities": [0.00825353991240263, 0.9539794921875, 0.023882411420345306, 0.002517187502235174, 0.0053288862109184265, 0.006038418505340815], "decide_seconds": 0.16115862399965408, "probabilities_f64_minus_runtime_max_abs": 6.372647831653921e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.009434123401746491, "ibn_sina": 0.0025391598670406156, "hypatia": 0.012113654232011111, "einstein": 0.00886253875979496, "sagan": 0.9623067485825342, "darwin": 0.0047437751568727695}, "confidence": 0.8789609916601411, "choice": "sagan"}}, "usage": {"input_tokens": 293, "output_tokens": 1}, "latency_ms": 164.71372599698952, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 293, "compiled_prompt_sha256": "78775807212758e40edf769f9eb7a399a8186be9fa4466e97223edc44ad77b5e", "compiled_ids_sha256": "7cec2fc6c5f8ca6a7e31daf6344b14e7480672c983adbefea1ea42e028662f59", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 13.5625, 15.125, 14.8125, 19.5, 14.1875], "runtime_probabilities": [0.009434124454855919, 0.0025391599629074335, 0.012113654986023903, 0.008862538263201714, 0.9623067378997803, 0.004743775352835655], "decide_seconds": 0.1612340329993458, "probabilities_f64_minus_runtime_max_abs": 1.0682753903701325e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:25Z"} {"id": "c-sagan-01-starter-03", "uid": "c088", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "prompt_sha256_by_order": ["ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f"], "request_sha256": "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "compiled_prompt_sha256_by_order": ["3824b179b4488ce489b4ef53aaa0bc5aac8823a7a8d88bcb5a7f0cf96ad47e1b", "dbc05fef85bcc0d56f0a06133eb4389f636ed6460a3f4b3cf3d4405cbf252d2a", "e452353fa315daca4512110c64c8bab66fbc44387cf2c7967e7b40996297fed1", "c668d5cb4b9609114af6406ec50b06bc347bf14cd986105eb07aab8d3ce32c19", "224425416d064d538ee268e1d83ab285dc072fbc957d6598f7ce41c93e35e9cf", "d761f71fff8dabb20c919c88489e7ab111f0c5dc12bd69b7aa2258f047408de5"], "compiled_tokens_by_order": [302, 302, 302, 302, 302, 302], "choice": "sagan", "probs": {"darwin": 0.13222605583698638, "einstein": 0.033431982512679474, "hypatia": 0.013936523752092138, "ibn_sina": 0.010853775611422591, "sagan": 0.7609084147572049, "socrates": 0.04864324752961447}, "confidence": 0.5285324807563957, "seconds": 0.16523855399949147, "seconds_all_orders": 0.9914313239969488, "seconds_by_order": [0.16527159199904418, 0.16507967299912707, 0.16542427099921042, 0.16536089200235438, 0.16517612199822906, 0.16511877399898367], "argmax_stable": false, "spread": {"darwin": 0.7890236321063562, "einstein": 0.08089234276720016, "hypatia": 0.026007900496090954, "ibn_sina": 0.006822245104237877, "sagan": 0.746919977127798, "socrates": 0.06936443297328415}, "spread_2dp": {"darwin": 0.79, "einstein": 0.08, "hypatia": 0.019999999999999997, "ibn_sina": 0.01, "sagan": 0.75, "socrates": 0.07}, "choices_by_order": ["sagan", "darwin", "darwin", "sagan", "darwin", "darwin"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.13222605583698638, "einstein": 0.033431982512679474, "hypatia": 0.013936523752092138, "ibn_sina": 0.010853775611422591, "sagan": 0.7609084147572049, "socrates": 0.04864324752961447}, {"ibn_sina": 0.007189663133562159, "hypatia": 0.007653356567162933, "sagan": 0.053124824226304324, "socrates": 0.013432059084576987, "darwin": 0.8310120692842803, "einstein": 0.08758802770411342}, {"hypatia": 0.008597429641129885, "darwin": 0.876961664778648, "sagan": 0.08157014130671905, "ibn_sina": 0.005550917778862992, "einstein": 0.006695684936913263, "socrates": 0.020624161557726894}, {"ibn_sina": 0.007486264155172671, "einstein": 0.0079690867111767, "darwin": 0.16006338538142376, "sagan": 0.7173543248252364, "hypatia": 0.03151834935431066, "socrates": 0.07560858957267967}, {"einstein": 0.012344752879949386, "sagan": 0.013988437629406898, "socrates": 0.02306303466346523, "ibn_sina": 0.010234159594219397, "darwin": 0.9212496879433426, "hypatia": 0.019119927289616583}, {"socrates": 0.006244156599395518, "ibn_sina": 0.004031530507184714, "hypatia": 0.005510448858219706, "einstein": 0.009671138930560054, "sagan": 0.10397454216682751, "darwin": 0.8705681829378125}], "prompt_tokens": 302, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.13222605583698638, "einstein": 0.033431982512679474, "hypatia": 0.013936523752092138, "ibn_sina": 0.010853775611422591, "sagan": 0.7609084147572049, "socrates": 0.04864324752961447}, "confidence": 0.5285324807563957, "choice": "sagan"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.24668199781445, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "3824b179b4488ce489b4ef53aaa0bc5aac8823a7a8d88bcb5a7f0cf96ad47e1b", "compiled_ids_sha256": "4010bc1586d12cb94c392181608c684fece802cc25b6f26e7c35ffddfc2a80eb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 16.0, 15.125, 14.875, 19.125, 16.375], "runtime_probabilities": [0.13222603499889374, 0.03343198075890541, 0.013936523348093033, 0.010853774845600128, 0.7609084248542786, 0.04864324629306793], "decide_seconds": 0.16147889100102475, "probabilities_f64_minus_runtime_max_abs": 2.0838092640307337e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.007189663133562159, "hypatia": 0.007653356567162933, "sagan": 0.053124824226304324, "socrates": 0.013432059084576987, "darwin": 0.8310120692842803, "einstein": 0.08758802770411342}, "confidence": 0.6351571998782474, "choice": "darwin"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.05632400003378, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "dbc05fef85bcc0d56f0a06133eb4389f636ed6460a3f4b3cf3d4405cbf252d2a", "compiled_ids_sha256": "30493d5211dd4b74c6a5dff1ceb8a027ea94393458146b98e541f5c660849df8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.8125, 16.75, 15.375, 19.5, 17.25], "runtime_probabilities": [0.007189663127064705, 0.007653356064110994, 0.05312482267618179, 0.01343205850571394, 0.8310120701789856, 0.08758802711963654], "decide_seconds": 0.1614457309988211, "probabilities_f64_minus_runtime_max_abs": 1.55012253033604e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.008597429641129885, "darwin": 0.876961664778648, "sagan": 0.08157014130671905, "ibn_sina": 0.005550917778862992, "einstein": 0.006695684936913263, "socrates": 0.020624161557726894}, "confidence": 0.7193441436595126, "choice": "darwin"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.40154100221116, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "e452353fa315daca4512110c64c8bab66fbc44387cf2c7967e7b40996297fed1", "compiled_ids_sha256": "e6f4c5e9a68bd604e43ed9d7e0f62d833827757f30e8b18d03ceb075daf7654b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 19.5, 17.125, 14.4375, 14.625, 15.75], "runtime_probabilities": [0.008597430773079395, 0.8769617080688477, 0.08157014101743698, 0.00555091816931963, 0.006695684976875782, 0.02062416262924671], "decide_seconds": 0.16179680899949744, "probabilities_f64_minus_runtime_max_abs": 4.329019964721681e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.007486264155172671, "einstein": 0.0079690867111767, "darwin": 0.16006338538142376, "sagan": 0.7173543248252364, "hypatia": 0.03151834935431066, "socrates": 0.07560858957267967}, "confidence": 0.491610556015016, "choice": "sagan"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.33725199769833, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "c668d5cb4b9609114af6406ec50b06bc347bf14cd986105eb07aab8d3ce32c19", "compiled_ids_sha256": "ca2d8c3ec6069b674966e5365e7616dc00910813aeb55ee7c22d1291bfec4663", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.75, 17.75, 19.25, 16.125, 17.0], "runtime_probabilities": [0.0074862646870315075, 0.007969086989760399, 0.16006338596343994, 0.7173542976379395, 0.031518351286649704, 0.07560858875513077], "decide_seconds": 0.16154447099688696, "probabilities_f64_minus_runtime_max_abs": 2.7187296969266583e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.012344752879949386, "sagan": 0.013988437629406898, "socrates": 0.02306303466346523, "ibn_sina": 0.010234159594219397, "darwin": 0.9212496879433426, "hypatia": 0.019119927289616583}, "confidence": 0.7772994159238077, "choice": "darwin"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.15344299841672, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "224425416d064d538ee268e1d83ab285dc072fbc957d6598f7ce41c93e35e9cf", "compiled_ids_sha256": "26ba3d136f4edd057f6232d7696f6db1b956446a066809894f399bd60521bd41", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.9375, 15.0625, 15.5625, 14.75, 19.25, 15.375], "runtime_probabilities": [0.012344753369688988, 0.013988438062369823, 0.023063037544488907, 0.010234160348773003, 0.9212497472763062, 0.019119927659630775], "decide_seconds": 0.16153412099811248, "probabilities_f64_minus_runtime_max_abs": 5.9332963542324535e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.006244156599395518, "ibn_sina": 0.004031530507184714, "hypatia": 0.005510448858219706, "einstein": 0.009671138930560054, "sagan": 0.10397454216682751, "darwin": 0.8705681829378125}, "confidence": 0.7301692111642202, "choice": "darwin"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.09587399923475, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "d761f71fff8dabb20c919c88489e7ab111f0c5dc12bd69b7aa2258f047408de5", "compiled_ids_sha256": "ecb3959f6e14b31a8b1e7778935fab75c3e03de4dd2f588aeb0cb8a15c1cdea9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.875, 14.1875, 14.75, 17.125, 19.25], "runtime_probabilities": [0.006244156509637833, 0.004031531047075987, 0.005510448943823576, 0.009671139530837536, 0.10397454351186752, 0.8705682158470154], "decide_seconds": 0.16151876100047957, "probabilities_f64_minus_runtime_max_abs": 3.290920291831867e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:26Z"} {"id": "c-sagan-01-starter-03", "uid": "c089", "task": "kperm", "label": "sagan", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "prompt_sha256_by_order": ["d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f"], "request_sha256": "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "compiled_prompt_sha256_by_order": ["ce02283a565951bd24fed4ecdd1be1ff580fd4e71e1ed42c0be57924ef7735f0", "ba4e9a9fb6557bd03d3c8dfc7d34437d16c0f3251f9ba2798b92a02599923bae", "137e9c0c7561ce673df16cf8f93803152e90d488fcb4915a3b14984231405acf", "52dedeb334558076d3d1b6cc3cd0492d986dda37e5fe9cdd5907369c6f87e4f3", "b77101de530c3776c0c34577a0fa1c2fc7bf98cdf89c29c7fa5752c979c8777a", "334b36f7502e98d9cde994d77acd7496f6f4d6800dec36889109609a6a234c98"], "compiled_tokens_by_order": [355, 355, 355, 355, 355, 355], "choice": "einstein", "probs": {"darwin": 0.018845622581294037, "einstein": 0.6240813092224897, "hypatia": 0.03989618331767567, "ibn_sina": 0.006118277869921341, "sagan": 0.20260953687131053, "socrates": 0.1084490701373087}, "confidence": 0.3898916374315142, "seconds": 0.17091976733415018, "seconds_all_orders": 1.0255186040049011, "seconds_by_order": [0.17106888800117304, 0.17073241999969468, 0.17064379100338556, 0.17139039500034414, 0.17087991000153124, 0.17080319999877247], "argmax_stable": true, "spread": {"darwin": 0.03541623285057553, "einstein": 0.2434196953268546, "hypatia": 0.06697908749454737, "ibn_sina": 0.016165165473886932, "sagan": 0.3377479886352596, "socrates": 0.13922502603477085}, "spread_2dp": {"darwin": 0.030000000000000002, "einstein": 0.24, "hypatia": 0.06999999999999999, "ibn_sina": 0.01, "sagan": 0.32999999999999996, "socrates": 0.13}, "choices_by_order": ["einstein", "einstein", "einstein", "einstein", "einstein", "einstein"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.018845622581294037, "einstein": 0.6240813092224897, "hypatia": 0.03989618331767567, "ibn_sina": 0.006118277869921341, "sagan": 0.20260953687131053, "socrates": 0.1084490701373087}, {"ibn_sina": 0.020542270544128858, "hypatia": 0.029888827278471856, "sagan": 0.08648618938903646, "socrates": 0.08648618938903646, "darwin": 0.05245652550616404, "einstein": 0.7241399978931623}, {"hypatia": 0.025782096117100643, "darwin": 0.0375127285012427, "sagan": 0.1483657193085043, "ibn_sina": 0.007863095056968005, "einstein": 0.664929022637769, "socrates": 0.11554733837841542}, {"ibn_sina": 0.022283443343808274, "einstein": 0.6117623877202111, "darwin": 0.030457855822313838, "sagan": 0.1365026395447965, "hypatia": 0.04431591872914562, "socrates": 0.15467775483972473}, {"einstein": 0.661044726095316, "sagan": 0.11487234885070242, "socrates": 0.0696736015311563, "ibn_sina": 0.010684792852777231, "darwin": 0.054261855431869566, "hypatia": 0.08946267523817865}, {"socrates": 0.015452728804953887, "ibn_sina": 0.006441655739177341, "hypatia": 0.022483587743631284, "einstein": 0.4807203025663077, "sagan": 0.4242341780242961, "darwin": 0.050667547121633634}], "prompt_tokens": 355, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.018845622581294037, "einstein": 0.6240813092224897, "hypatia": 0.03989618331767567, "ibn_sina": 0.006118277869921341, "sagan": 0.20260953687131053, "socrates": 0.1084490701373087}, "confidence": 0.3898916374315142, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.03923799732001, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "ce02283a565951bd24fed4ecdd1be1ff580fd4e71e1ed42c0be57924ef7735f0", "compiled_ids_sha256": "fcc59376315f837186e1531a6bc1a77307e5878970fb6cc3e97bcfa34479b565", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 19.0, 16.25, 14.375, 17.875, 17.25], "runtime_probabilities": [0.01884562335908413, 0.6240813136100769, 0.03989618271589279, 0.006118278484791517, 0.20260953903198242, 0.1084490641951561], "decide_seconds": 0.16674648099797196, "probabilities_f64_minus_runtime_max_abs": 5.942152603299888e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.020542270544128858, "hypatia": 0.029888827278471856, "sagan": 0.08648618938903646, "socrates": 0.08648618938903646, "darwin": 0.05245652550616404, "einstein": 0.7241399978931623}, "confidence": 0.44384962643244685, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.70582999804174, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "ba4e9a9fb6557bd03d3c8dfc7d34437d16c0f3251f9ba2798b92a02599923bae", "compiled_ids_sha256": "1f5cada7fea9a22086a3299315822a482ee78493540d6047321546300c4ed697", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.9375, 17.0, 17.0, 16.5, 19.125], "runtime_probabilities": [0.020542271435260773, 0.029888829216361046, 0.08648619055747986, 0.08648619055747986, 0.05245652794837952, 0.7241399884223938], "decide_seconds": 0.16658461300175986, "probabilities_f64_minus_runtime_max_abs": 9.470768502772842e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.025782096117100643, "darwin": 0.0375127285012427, "sagan": 0.1483657193085043, "ibn_sina": 0.007863095056968005, "einstein": 0.664929022637769, "socrates": 0.11554733837841542}, "confidence": 0.4087572537755233, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.6183609967411, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "137e9c0c7561ce673df16cf8f93803152e90d488fcb4915a3b14984231405acf", "compiled_ids_sha256": "e3e91a6cf40aacb212487a26c4dfef10188da63f6c7e0c252e48dadc5f4cbba3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 16.375, 17.75, 14.8125, 19.25, 17.5], "runtime_probabilities": [0.025782093405723572, 0.037512727081775665, 0.14836570620536804, 0.007863094098865986, 0.6649289727210999, 0.115547314286232], "decide_seconds": 0.16647466300128144, "probabilities_f64_minus_runtime_max_abs": 4.991666913145565e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.022283443343808274, "einstein": 0.6117623877202111, "darwin": 0.030457855822313838, "sagan": 0.1365026395447965, "hypatia": 0.04431591872914562, "socrates": 0.15467775483972473}, "confidence": 0.33564522465644064, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.3641460009967, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "52dedeb334558076d3d1b6cc3cd0492d986dda37e5fe9cdd5907369c6f87e4f3", "compiled_ids_sha256": "1eba59e423a50ec4dac2773808a644e078beb3ac9d36000edd522f9d8af4c8ce", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 19.0, 16.0, 17.5, 16.375, 17.625], "runtime_probabilities": [0.022283444181084633, 0.6117623448371887, 0.03045785427093506, 0.13650263845920563, 0.044315919280052185, 0.15467774868011475], "decide_seconds": 0.16718407699954696, "probabilities_f64_minus_runtime_max_abs": 4.288302235533337e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.661044726095316, "sagan": 0.11487234885070242, "socrates": 0.0696736015311563, "ibn_sina": 0.010684792852777231, "darwin": 0.054261855431869566, "hypatia": 0.08946267523817865}, "confidence": 0.3691224910201186, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.85251999742468, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "b77101de530c3776c0c34577a0fa1c2fc7bf98cdf89c29c7fa5752c979c8777a", "compiled_ids_sha256": "4b675927e7bc281b123d8a453579b681e3f71b80fb3ce354b839ff51b9fe3b7c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 17.375, 16.875, 15.0, 16.625, 17.125], "runtime_probabilities": [0.661044716835022, 0.11487233638763428, 0.06967359781265259, 0.010684792883694172, 0.05426185578107834, 0.08946267515420914], "decide_seconds": 0.16655690200059325, "probabilities_f64_minus_runtime_max_abs": 1.246306814695064e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.015452728804953887, "ibn_sina": 0.006441655739177341, "hypatia": 0.022483587743631284, "einstein": 0.4807203025663077, "sagan": 0.4242341780242961, "darwin": 0.050667547121633634}, "confidence": 0.4143994637906627, "choice": "einstein"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.7761899997422, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "334b36f7502e98d9cde994d77acd7496f6f4d6800dec36889109609a6a234c98", "compiled_ids_sha256": "bfb720ffd6b7eb966c39f15507c690820bb76fb08be08c40507efe7ccd9bd5e5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 14.5625, 15.8125, 18.875, 18.75, 16.625], "runtime_probabilities": [0.015452730469405651, 0.006441656034439802, 0.022483589127659798, 0.48072031140327454, 0.42423418164253235, 0.050667546689510345], "decide_seconds": 0.16665713200200116, "probabilities_f64_minus_runtime_max_abs": 8.836966824965486e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:27Z"} {"id": "c-socrates-01-dessert-01", "uid": "c090", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "prompt_sha256_by_order": ["b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee"], "request_sha256": "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "compiled_prompt_sha256_by_order": ["e5874d1b8d5df9f9ce009834e203a80ed9bfc65e1e117e65b805719bfafff002", "826cca657c25104c48f94c082d5d94b4889b76d32f854dc1bd230a522523032f", "c604dc6c278592df989e225e1321c8e7d15f9b05fd35a13b0ec2f4697065d631", "3050f73463d83ddbc40ca354099fb692e24640d71af15c37a17ce5bb9d2d6274", "99440821fae71a77847677cf8db901ebc7bdf278b18cebc2be67df818e7af741", "8967dc6aabe1738656ebdcacb7913cbd49fb8eb32772829f8826cceafde8b0f8"], "compiled_tokens_by_order": [409, 409, 409, 409, 409, 409], "choice": "socrates", "probs": {"darwin": 0.0032923602593792427, "einstein": 0.0022628039081888183, "hypatia": 0.004227474253935593, "ibn_sina": 0.006969926723800767, "sagan": 0.011491466444751582, "socrates": 0.9717559684099439}, "confidence": 0.9054067218663747, "seconds": 0.17742123733478365, "seconds_all_orders": 1.064527424008702, "seconds_by_order": [0.17756286800067755, 0.17738402000031783, 0.1772958200017456, 0.1776592870010063, 0.177409219002584, 0.17721621000237064], "argmax_stable": true, "spread": {"darwin": 0.00666231967920632, "einstein": 0.005218892173225418, "hypatia": 0.03146734205748036, "ibn_sina": 0.027356044478241283, "sagan": 0.023829044876456998, "socrates": 0.06494696302623637}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.04, "ibn_sina": 0.03, "sagan": 0.03, "socrates": 0.05999999999999994}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0032923602593792427, "einstein": 0.0022628039081888183, "hypatia": 0.004227474253935593, "ibn_sina": 0.006969926723800767, "sagan": 0.011491466444751582, "socrates": 0.9717559684099439}, {"ibn_sina": 0.0083446213290232, "hypatia": 0.013757954681098522, "sagan": 0.005061268679744554, "socrates": 0.9645070859654563, "darwin": 0.005061268679744554, "einstein": 0.0032678006649327134}, {"hypatia": 0.007443579984544116, "darwin": 0.0031029457859955804, "sagan": 0.0033030685955148173, "ibn_sina": 0.006992596271577673, "einstein": 0.004241224029704106, "socrates": 0.9749165853326638}, {"ibn_sina": 0.004011775206516459, "einstein": 0.0017802144565416476, "darwin": 0.0017802144565416476, "sagan": 0.0022858406093544305, "hypatia": 0.00849292817884166, "socrates": 0.9816490270922043}, {"einstein": 0.0069991066297670655, "sagan": 0.005120655799001642, "socrates": 0.9167020640659679, "ibn_sina": 0.03136781968475774, "darwin": 0.008442534135747968, "hypatia": 0.03136781968475774}, {"socrates": 0.9205814459531849, "ibn_sina": 0.010226736096264028, "hypatia": 0.03569481631141595, "einstein": 0.0031189782895159335, "sagan": 0.02611488548581143, "darwin": 0.004263137863807727}], "prompt_tokens": 409, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0032923602593792427, "einstein": 0.0022628039081888183, "hypatia": 0.004227474253935593, "ibn_sina": 0.006969926723800767, "sagan": 0.011491466444751582, "socrates": 0.9717559684099439}, "confidence": 0.9054067218663747, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.53322899807245, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "e5874d1b8d5df9f9ce009834e203a80ed9bfc65e1e117e65b805719bfafff002", "compiled_ids_sha256": "2fbf7e93347627cacd15188db8ccc02538cff99269ff3adb9d0ea548870e9cb9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.8125, 13.4375, 14.0625, 14.5625, 15.0625, 19.5], "runtime_probabilities": [0.0032923605758696795, 0.0022628039587289095, 0.004227474331855774, 0.006969926413148642, 0.01149146631360054, 0.9717560410499573], "decide_seconds": 0.17261709599915775, "probabilities_f64_minus_runtime_max_abs": 7.264001333773251e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0083446213290232, "hypatia": 0.013757954681098522, "sagan": 0.005061268679744554, "socrates": 0.9645070859654563, "darwin": 0.005061268679744554, "einstein": 0.0032678006649327134}, "confidence": 0.8850429984875546, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.35750999781885, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "826cca657c25104c48f94c082d5d94b4889b76d32f854dc1bd230a522523032f", "compiled_ids_sha256": "193f094f387d5b91ff8d61100935c73c288a7b16a6fe414732858b803e68dfc7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 15.375, 14.375, 19.625, 14.375, 13.9375], "runtime_probabilities": [0.008344621397554874, 0.01375795528292656, 0.0050612688064575195, 0.9645071625709534, 0.0050612688064575195, 0.0032678009010851383], "decide_seconds": 0.17268849500032957, "probabilities_f64_minus_runtime_max_abs": 7.6605497056903e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.007443579984544116, "darwin": 0.0031029457859955804, "sagan": 0.0033030685955148173, "ibn_sina": 0.006992596271577673, "einstein": 0.004241224029704106, "socrates": 0.9749165853326638}, "confidence": 0.9129868931012945, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.2705599978508, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "c604dc6c278592df989e225e1321c8e7d15f9b05fd35a13b0ec2f4697065d631", "compiled_ids_sha256": "cae13c3737b774dd07b3d009ba5fc432b7b06ce6538c5c9ce26494ea03bb8699", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 13.875, 13.9375, 14.6875, 14.1875, 19.625], "runtime_probabilities": [0.007443580310791731, 0.0031029460951685905, 0.0033030689228326082, 0.006992596201598644, 0.004241224378347397, 0.9749166369438171], "decide_seconds": 0.1726191160014423, "probabilities_f64_minus_runtime_max_abs": 5.1611153351061034e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004011775206516459, "einstein": 0.0017802144565416476, "darwin": 0.0017802144565416476, "sagan": 0.0022858406093544305, "hypatia": 0.00849292817884166, "socrates": 0.9816490270922043}, "confidence": 0.9345555408304022, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.63346699939575, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "3050f73463d83ddbc40ca354099fb692e24640d71af15c37a17ce5bb9d2d6274", "compiled_ids_sha256": "b8d5c80860c4fcf8751da213e0ce5b2ccf8cde7751880e1e472956126126543c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 13.1875, 13.1875, 13.4375, 14.75, 19.5], "runtime_probabilities": [0.004011775366961956, 0.0017802143702283502, 0.0017802143702283502, 0.002285840455442667, 0.008492927998304367, 0.9816489815711975], "decide_seconds": 0.1728930639983446, "probabilities_f64_minus_runtime_max_abs": 4.5521006741822134e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0069991066297670655, "sagan": 0.005120655799001642, "socrates": 0.9167020640659679, "ibn_sina": 0.03136781968475774, "darwin": 0.008442534135747968, "hypatia": 0.03136781968475774}, "confidence": 0.7773339682846797, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.3829490011849, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "99440821fae71a77847677cf8db901ebc7bdf278b18cebc2be67df818e7af741", "compiled_ids_sha256": "4f9249d1be322fdd961969450b842cde09ad594b0ff5d94e319d6d7b6fc23857", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.75, 14.4375, 19.625, 16.25, 14.9375, 16.25], "runtime_probabilities": [0.006999106612056494, 0.005120656453073025, 0.9167020916938782, 0.031367819756269455, 0.008442534133791924, 0.031367819756269455], "decide_seconds": 0.17256490599902463, "probabilities_f64_minus_runtime_max_abs": 2.762791029287115e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9205814459531849, "ibn_sina": 0.010226736096264028, "hypatia": 0.03569481631141595, "einstein": 0.0031189782895159335, "sagan": 0.02611488548581143, "darwin": 0.004263137863807727}, "confidence": 0.7887740466127879, "choice": "socrates"}}, "usage": {"input_tokens": 409, "output_tokens": 1}, "latency_ms": 177.19115099680494, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 409, "compiled_prompt_sha256": "8967dc6aabe1738656ebdcacb7913cbd49fb8eb32772829f8826cceafde8b0f8", "compiled_ids_sha256": "32c12ebbe401fa75f69eef505c22f6e7bd7769a0a47e74220ed78170cf127da8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.375, 14.875, 16.125, 13.6875, 15.8125, 14.0], "runtime_probabilities": [0.9205814599990845, 0.010226736776530743, 0.03569481521844864, 0.0031189783476293087, 0.02611488662660122, 0.004263138398528099], "decide_seconds": 0.17260997599805705, "probabilities_f64_minus_runtime_max_abs": 1.4045899554560037e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:28Z"} {"id": "c-socrates-01-dessert-01", "uid": "c091", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "prompt_sha256_by_order": ["42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512"], "request_sha256": "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "compiled_prompt_sha256_by_order": ["38c5218cbd49d0ad53554b844ea9581c36f8d9f6dab540ad7633c7aefaaec7f9", "8f189933761c184a074abc08bc8bdc517ccc27b4dd31089bffe4818595ededd7", "3828f5a05800bfa57043bb88653ff9a2be72492389e3b064f30b37f499644d4b", "95574326175c1fb3153918b7d4aefc7008232a67ab2aa789172cef901c60d16e", "9e178a50169ff30850f5576659d9957fd1fad665dfdbf39606bb52e71b42bc9a", "70a83edfe0e32d7f12da55e8b50a5b267829b2bde52579d104e353a309346941"], "compiled_tokens_by_order": [291, 291, 291, 291, 291, 291], "choice": "socrates", "probs": {"darwin": 0.005460495944458775, "einstein": 0.0029227928596531564, "hypatia": 0.00331195820755074, "ibn_sina": 0.0037529403195081003, "sagan": 0.007011415580405401, "socrates": 0.9775403970884239}, "confidence": 0.9205462134028627, "seconds": 0.16501323916721353, "seconds_all_orders": 0.9900794350032811, "seconds_by_order": [0.16525439199904213, 0.1648955950004165, 0.16485622599793714, 0.16531926200332236, 0.16492932500113966, 0.1648246350014233], "argmax_stable": true, "spread": {"darwin": 0.005248287995613173, "einstein": 0.005473217089493369, "hypatia": 0.0028880571779323063, "ibn_sina": 0.005177111086781042, "sagan": 0.004702467708703877, "socrates": 0.019064273577468205}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.0, "ibn_sina": 0.01, "sagan": 0.01, "socrates": 0.020000000000000018}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.005460495944458775, "einstein": 0.0029227928596531564, "hypatia": 0.00331195820755074, "ibn_sina": 0.0037529403195081003, "sagan": 0.007011415580405401, "socrates": 0.9775403970884239}, {"ibn_sina": 0.006975321497333242, "hypatia": 0.004230746349485245, "sagan": 0.0039744183861568656, "socrates": 0.972508114535168, "darwin": 0.006155699615928329, "einstein": 0.006155699615928329}, {"hypatia": 0.0045488533625603796, "darwin": 0.002759019030929778, "sagan": 0.0035426505608390168, "ibn_sina": 0.0035426505608390168, "einstein": 0.0033280122138356584, "socrates": 0.9822788142709961}, {"ibn_sina": 0.0033422113634546894, "einstein": 0.002157893841825076, "darwin": 0.0020271536631752325, "sagan": 0.0024452140687465157, "hypatia": 0.0035577654769298208, "socrates": 0.9864697615858687}, {"einstein": 0.00656388683610772, "sagan": 0.0031005605973543292, "socrates": 0.9741671813385632, "ibn_sina": 0.004802242195933532, "darwin": 0.00656388683610772, "hypatia": 0.004802242195933532}, {"socrates": 0.9915723881126361, "ibn_sina": 0.0017982104105521998, "hypatia": 0.0019141850180012258, "einstein": 0.0010906697466143511, "sagan": 0.0023089478717015235, "darwin": 0.0013155988404945466}], "prompt_tokens": 291, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.005460495944458775, "einstein": 0.0029227928596531564, "hypatia": 0.00331195820755074, "ibn_sina": 0.0037529403195081003, "sagan": 0.007011415580405401, "socrates": 0.9775403970884239}, "confidence": 0.9205462134028627, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 165.22917200200027, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "38c5218cbd49d0ad53554b844ea9581c36f8d9f6dab540ad7633c7aefaaec7f9", "compiled_ids_sha256": "2b8caa07c5125e82419f0a570ae0c7c4b742c0c93eaa8a8cd7f945e8e8e0643e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.1875, 13.5625, 13.6875, 13.8125, 14.4375, 19.375], "runtime_probabilities": [0.0054604965262115, 0.0029227931518107653, 0.003311958396807313, 0.0037529405672103167, 0.007011415436863899, 0.9775404334068298], "decide_seconds": 0.16159866999805672, "probabilities_f64_minus_runtime_max_abs": 3.631840594930935e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.006975321497333242, "hypatia": 0.004230746349485245, "sagan": 0.0039744183861568656, "socrates": 0.972508114535168, "darwin": 0.006155699615928329, "einstein": 0.006155699615928329}, "confidence": 0.9053958104622768, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 164.87328500079457, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "8f189933761c184a074abc08bc8bdc517ccc27b4dd31089bffe4818595ededd7", "compiled_ids_sha256": "f93e85791447fcb0c6ec05b6330a6b949d655214d143738ba2cac876e044a327", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.1875, 14.125, 19.625, 14.5625, 14.5625], "runtime_probabilities": [0.006975321099162102, 0.004230746533721685, 0.003974418621510267, 0.9725081324577332, 0.0061556994915008545, 0.0061556994915008545], "decide_seconds": 0.16137589199934155, "probabilities_f64_minus_runtime_max_abs": 1.7922565209715913e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.0045488533625603796, "darwin": 0.002759019030929778, "sagan": 0.0035426505608390168, "ibn_sina": 0.0035426505608390168, "einstein": 0.0033280122138356584, "socrates": 0.9822788142709961}, "confidence": 0.9345211788031876, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 164.83403600068414, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "3828f5a05800bfa57043bb88653ff9a2be72492389e3b064f30b37f499644d4b", "compiled_ids_sha256": "8b6092398dd4000f63708de61cb52a18bd7834aee196cd8c7b3f0257c113659b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.25, 13.75, 14.0, 14.0, 13.9375, 19.625], "runtime_probabilities": [0.004548853728920221, 0.0027590191457420588, 0.0035426507238298655, 0.0035426507238298655, 0.003328012302517891, 0.9822788238525391], "decide_seconds": 0.1613709220000601, "probabilities_f64_minus_runtime_max_abs": 9.581543003456261e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0033422113634546894, "einstein": 0.002157893841825076, "darwin": 0.0020271536631752325, "sagan": 0.0024452140687465157, "hypatia": 0.0035577654769298208, "socrates": 0.9864697615858687}, "confidence": 0.9480537074151043, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 165.29759200057015, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "95574326175c1fb3153918b7d4aefc7008232a67ab2aa789172cef901c60d16e", "compiled_ids_sha256": "128d2c2333aeb543374911fd68ac206b72525444319b8b78b7d1593ad2c78666", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.6875, 13.25, 13.1875, 13.375, 13.75, 19.375], "runtime_probabilities": [0.003342211712151766, 0.0021578939631581306, 0.0020271537359803915, 0.0024452144280076027, 0.003557765856385231, 0.9864698052406311], "decide_seconds": 0.16177483900173684, "probabilities_f64_minus_runtime_max_abs": 4.3654762360567645e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.00656388683610772, "sagan": 0.0031005605973543292, "socrates": 0.9741671813385632, "ibn_sina": 0.004802242195933532, "darwin": 0.00656388683610772, "hypatia": 0.004802242195933532}, "confidence": 0.9103321266772083, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 164.90563499974087, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "9e178a50169ff30850f5576659d9957fd1fad665dfdbf39606bb52e71b42bc9a", "compiled_ids_sha256": "31a94af66f20d8fb8bc8b4405817e0492d805ffff7fa1acef161bbe9042c2243", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 13.75, 19.5, 14.1875, 14.5, 14.1875], "runtime_probabilities": [0.006563886534422636, 0.0031005607452243567, 0.9741671681404114, 0.004802241921424866, 0.006563886534422636, 0.004802241921424866], "decide_seconds": 0.161224913001206, "probabilities_f64_minus_runtime_max_abs": 1.3198151793680779e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9915723881126361, "ibn_sina": 0.0017982104105521998, "hypatia": 0.0019141850180012258, "einstein": 0.0010906697466143511, "sagan": 0.0023089478717015235, "darwin": 0.0013155988404945466}, "confidence": 0.9654405560106273, "choice": "socrates"}}, "usage": {"input_tokens": 291, "output_tokens": 1}, "latency_ms": 164.8018560008495, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 291, "compiled_prompt_sha256": "70a83edfe0e32d7f12da55e8b50a5b267829b2bde52579d104e353a309346941", "compiled_ids_sha256": "62dd4eda8cbfe25d72fdf9049c9a989998478590190a0c47e033df08a312ecc9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 13.1875, 13.25, 12.6875, 13.4375, 12.875], "runtime_probabilities": [0.9915724992752075, 0.0017982105491682887, 0.0019141852390021086, 0.001090669771656394, 0.002308947965502739, 0.001315598958171904], "decide_seconds": 0.16129764300058014, "probabilities_f64_minus_runtime_max_abs": 1.1116257137011587e-07}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:29Z"} {"id": "c-socrates-01-dessert-01", "uid": "c092", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "prompt_sha256_by_order": ["67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261"], "request_sha256": "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "compiled_prompt_sha256_by_order": ["9a53e19f51ada7e3ca7748e01853f6e0dfe0321fc114aca34ef2b3bce186324a", "de8d7cb4024a379a66d860fe4bf7cdef4bcd513d536e55204b94c41028771cc6", "51f75317bdb836de7c8910f78a3432898acf6b93f46c3428c37789edc7151bf3", "f84157b0f20df3bff51dcaa2026f36e96c6e497703ac65f24ff3bfd71fedcffc", "7ba923ff5a29def4080b4cc3b74adbc94ad27b95cd6f272861e5486c94febab7", "cddcd9cbdf38b13843bf7a9aaff4de5fd4e017ae4dcc0dae94ea85c961efcdc9"], "compiled_tokens_by_order": [361, 361, 361, 361, 361, 361], "choice": "socrates", "probs": {"darwin": 0.04867316682228247, "einstein": 0.0229915760228299, "hypatia": 0.16988604503551896, "ibn_sina": 0.07081903984870049, "sagan": 0.28009473604518004, "socrates": 0.4075354362254882}, "confidence": 0.19365416477292718, "seconds": 0.17104020449975602, "seconds_all_orders": 1.0262412269985361, "seconds_by_order": [0.1711742970001069, 0.17080426900065504, 0.17087355900002876, 0.17095392899864237, 0.17143480500089936, 0.17100036799820373], "argmax_stable": false, "spread": {"darwin": 0.11217642173215331, "einstein": 0.04833604610149614, "hypatia": 0.17510477549526124, "ibn_sina": 0.05969536689633475, "sagan": 0.3410830493817126, "socrates": 0.33561643278984454}, "spread_2dp": {"darwin": 0.11000000000000001, "einstein": 0.049999999999999996, "hypatia": 0.17, "ibn_sina": 0.060000000000000005, "sagan": 0.34, "socrates": 0.32999999999999996}, "choices_by_order": ["socrates", "sagan", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.04867316682228247, "einstein": 0.0229915760228299, "hypatia": 0.16988604503551896, "ibn_sina": 0.07081903984870049, "sagan": 0.28009473604518004, "socrates": 0.4075354362254882}, {"ibn_sina": 0.05334249935292518, "hypatia": 0.11292607201630418, "sagan": 0.34783678970675114, "socrates": 0.34783678970675114, "darwin": 0.07761287859278308, "einstein": 0.060444970624485346}, {"hypatia": 0.05757048791474189, "darwin": 0.02894825928272665, "sagan": 0.3312952838131292, "ibn_sina": 0.0652359093228922, "einstein": 0.034918266014906586, "socrates": 0.48203179365160337}, {"ibn_sina": 0.0372981133379881, "einstein": 0.01210892452298921, "darwin": 0.04789172552047677, "sagan": 0.31229327947952656, "hypatia": 0.18941544882653905, "socrates": 0.4009925083124804}, {"einstein": 0.03568190730587603, "sagan": 0.10990801208509073, "socrates": 0.3836166559498274, "ibn_sina": 0.09699348023432285, "darwin": 0.14112468101487996, "hypatia": 0.23267526341000314}, {"socrates": 0.14641536086175885, "ibn_sina": 0.06103500304079043, "hypatia": 0.21303309302208492, "einstein": 0.014497083915742318, "sagan": 0.4509910614668033, "darwin": 0.11402839769282011}], "prompt_tokens": 361, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.04867316682228247, "einstein": 0.0229915760228299, "hypatia": 0.16988604503551896, "ibn_sina": 0.07081903984870049, "sagan": 0.28009473604518004, "socrates": 0.4075354362254882}, "confidence": 0.19365416477292718, "choice": "socrates"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 171.14624699752312, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "9a53e19f51ada7e3ca7748e01853f6e0dfe0321fc114aca34ef2b3bce186324a", "compiled_ids_sha256": "08a973dfb69a66733142afb8c93b88554f14300d2455d7ab0777f84c253ecdf1", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 15.875, 17.875, 17.0, 18.375, 18.75], "runtime_probabilities": [0.04867316409945488, 0.02299157716333866, 0.1698860377073288, 0.07081902772188187, 0.2800947427749634, 0.4075354337692261], "decide_seconds": 0.1668137199994817, "probabilities_f64_minus_runtime_max_abs": 1.2126818618729551e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.05334249935292518, "hypatia": 0.11292607201630418, "sagan": 0.34783678970675114, "socrates": 0.34783678970675114, "darwin": 0.07761287859278308, "einstein": 0.060444970624485346}, "confidence": 0.15988782872099194, "choice": "sagan"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 170.77796000012313, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "de8d7cb4024a379a66d860fe4bf7cdef4bcd513d536e55204b94c41028771cc6", "compiled_ids_sha256": "aa9683662fd38d0c550a7d7ecc657147256be25d584da37e731d6af942259ddf", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 17.25, 18.375, 18.375, 16.875, 16.625], "runtime_probabilities": [0.05334250256419182, 0.11292607337236404, 0.34783679246902466, 0.34783679246902466, 0.07761287689208984, 0.060444965958595276], "decide_seconds": 0.1666365519995452, "probabilities_f64_minus_runtime_max_abs": 4.665890070143508e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.05757048791474189, "darwin": 0.02894825928272665, "sagan": 0.3312952838131292, "ibn_sina": 0.0652359093228922, "einstein": 0.034918266014906586, "socrates": 0.48203179365160337}, "confidence": 0.2856923646443006, "choice": "socrates"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 170.84977899867226, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "51f75317bdb836de7c8910f78a3432898acf6b93f46c3428c37789edc7151bf3", "compiled_ids_sha256": "de468230ede7b10401815cdcd6e89ca384dfc7693725852a4e468fc3779c9fe6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 15.9375, 18.375, 16.75, 16.125, 18.75], "runtime_probabilities": [0.05757048726081848, 0.028948256745934486, 0.33129528164863586, 0.06523590534925461, 0.03491826727986336, 0.48203176259994507], "decide_seconds": 0.16673279099995852, "probabilities_f64_minus_runtime_max_abs": 3.1051658300196294e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0372981133379881, "einstein": 0.01210892452298921, "darwin": 0.04789172552047677, "sagan": 0.31229327947952656, "hypatia": 0.18941544882653905, "socrates": 0.4009925083124804}, "confidence": 0.23724046890043926, "choice": "socrates"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 170.9286989971588, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "f84157b0f20df3bff51dcaa2026f36e96c6e497703ac65f24ff3bfd71fedcffc", "compiled_ids_sha256": "e79370f98d912912deab76fbcb1ee8c1917dfa2ed819315b041773760006fcb4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 15.125, 16.5, 18.375, 17.875, 18.625], "runtime_probabilities": [0.037298113107681274, 0.012108924798667431, 0.047891728579998016, 0.31229329109191895, 0.18941546976566315, 0.4009925127029419], "decide_seconds": 0.16673548100152402, "probabilities_f64_minus_runtime_max_abs": 2.0939124101282403e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.03568190730587603, "sagan": 0.10990801208509073, "socrates": 0.3836166559498274, "ibn_sina": 0.09699348023432285, "darwin": 0.14112468101487996, "hypatia": 0.23267526341000314}, "confidence": 0.12316946733166867, "choice": "socrates"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 171.40965499856975, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "7ba923ff5a29def4080b4cc3b74adbc94ad27b95cd6f272861e5486c94febab7", "compiled_ids_sha256": "6ed54a7b4bb717c0b75f1350027b93775f9e1466cf75e1c5351cc67606bc2db4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 17.375, 18.625, 17.25, 17.625, 18.125], "runtime_probabilities": [0.03568190336227417, 0.10990801453590393, 0.3836166262626648, 0.09699347615242004, 0.1411246955394745, 0.23267525434494019], "decide_seconds": 0.16720942800020566, "probabilities_f64_minus_runtime_max_abs": 2.9687162617086926e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.14641536086175885, "ibn_sina": 0.06103500304079043, "hypatia": 0.21303309302208492, "einstein": 0.014497083915742318, "sagan": 0.4509910614668033, "darwin": 0.11402839769282011}, "confidence": 0.19102275526411006, "choice": "sagan"}}, "usage": {"input_tokens": 361, "output_tokens": 1}, "latency_ms": 170.97599800035823, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 361, "compiled_prompt_sha256": "cddcd9cbdf38b13843bf7a9aaff4de5fd4e017ae4dcc0dae94ea85c961efcdc9", "compiled_ids_sha256": "c9b9f8ec626de520409705fd3ee7a2764748b93fd4a1736c06e601c13f97ab99", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.5, 16.625, 17.875, 15.1875, 18.625, 17.25], "runtime_probabilities": [0.1464153677225113, 0.06103500351309776, 0.21303310990333557, 0.01449708454310894, 0.45099106431007385, 0.11402840167284012], "decide_seconds": 0.16669139099758468, "probabilities_f64_minus_runtime_max_abs": 1.6881250647360346e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:30Z"} {"id": "c-socrates-01-dessert-02", "uid": "c093", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "prompt_sha256_by_order": ["14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e"], "request_sha256": "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "compiled_prompt_sha256_by_order": ["98bf941e9d49195d9f221459067344330d3de649eeabf47b96667525ca3aa283", "a6bdb550b05093a7263f98cc854cd521f95703cdc49a66b59cea5fe8b2d41ad9", "9bca8f1e371997ae4d4ee49bce13b6556ba34f80e746c6c72e2117d9fc507c3a", "feef5a6a106e90235b1f563bc78fdb009e737e32d309db116fc776e8272ecd4c", "61380b6f56927125bd98ea094e716be7e591d0ff8fbe658560560b004d22de75", "e59ecb5de50a329c375c493f61934017ab12c03525749ada6d1185884cb8f983"], "compiled_tokens_by_order": [343, 343, 343, 343, 343, 343], "choice": "sagan", "probs": {"darwin": 0.012367596332214221, "einstein": 0.010914365455595654, "hypatia": 0.03361861237157465, "ibn_sina": 0.014918168808544767, "sagan": 0.4640906285160354, "socrates": 0.4640906285160354}, "confidence": 0.4458146237139635, "seconds": 0.17073522850053754, "seconds_all_orders": 1.0244113710032252, "seconds_by_order": [0.17087513899969053, 0.17041556300318916, 0.17066567000074429, 0.17055095199975767, 0.1709492579975631, 0.17095478900228045], "argmax_stable": false, "spread": {"darwin": 0.15798484749729752, "einstein": 0.10192008668615182, "hypatia": 0.05299131634608045, "ibn_sina": 0.04263465610519719, "sagan": 0.7314234769500977, "socrates": 0.7139710027499331}, "spread_2dp": {"darwin": 0.16, "einstein": 0.1, "hypatia": 0.05, "ibn_sina": 0.04, "sagan": 0.73, "socrates": 0.72}, "choices_by_order": ["sagan", "socrates", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.012367596332214221, "einstein": 0.010914365455595654, "hypatia": 0.03361861237157465, "ibn_sina": 0.014918168808544767, "sagan": 0.4640906285160354, "socrates": 0.4640906285160354}, {"ibn_sina": 0.014345290773025756, "hypatia": 0.016255344048247952, "sagan": 0.06039593003954939, "socrates": 0.73577305295788, "darwin": 0.06039593003954939, "einstein": 0.11283445214174748}, {"hypatia": 0.02075878066761299, "darwin": 0.008129255312386804, "sagan": 0.09303440043158626, "ibn_sina": 0.0266548019966612, "einstein": 0.07245526390869803, "socrates": 0.7789674976830546}, {"ibn_sina": 0.015797622299323296, "einstein": 0.01230320061737927, "darwin": 0.017901051270612194, "sagan": 0.1498836672185462, "hypatia": 0.042942389629109914, "socrates": 0.7611720689650291}, {"einstein": 0.042000222611868684, "sagan": 0.08891447196706205, "socrates": 0.579795188876874, "ibn_sina": 0.0539293533401826, "darwin": 0.16611410280968433, "hypatia": 0.0692466603943284}, {"socrates": 0.06499649493312154, "ibn_sina": 0.011294697234985406, "hypatia": 0.05061932115081165, "einstein": 0.0573592054573351, "sagan": 0.791819406989647, "darwin": 0.02391087423409923}], "prompt_tokens": 343, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.012367596332214221, "einstein": 0.010914365455595654, "hypatia": 0.03361861237157465, "ibn_sina": 0.014918168808544767, "sagan": 0.4640906285160354, "socrates": 0.4640906285160354}, "confidence": 0.4458146237139635, "choice": "sagan"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.84904899820685, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "98bf941e9d49195d9f221459067344330d3de649eeabf47b96667525ca3aa283", "compiled_ids_sha256": "0f0f1ae75907c024b2d459061eabb496925526ca0d95c76d438b65fd51aa1e6e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.875, 16.0, 15.1875, 18.625, 18.625], "runtime_probabilities": [0.012367595918476582, 0.010914365760982037, 0.033618614077568054, 0.014918169938027859, 0.46409061551094055, 0.46409061551094055], "decide_seconds": 0.16673233099936624, "probabilities_f64_minus_runtime_max_abs": 1.3005094834372954e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.014345290773025756, "hypatia": 0.016255344048247952, "sagan": 0.06039593003954939, "socrates": 0.73577305295788, "darwin": 0.06039593003954939, "einstein": 0.11283445214174748}, "confidence": 0.4760257135777044, "choice": "socrates"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.39084299904061, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "a6bdb550b05093a7263f98cc854cd521f95703cdc49a66b59cea5fe8b2d41ad9", "compiled_ids_sha256": "60f0d47e1eef9297cc7b2f50cdbff5a6bf7dc1cea5a9ae9f3a1b2dddd8c91f0d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 15.3125, 16.625, 19.125, 16.625, 17.25], "runtime_probabilities": [0.014345292933285236, 0.016255345195531845, 0.06039593368768692, 0.7357730865478516, 0.06039593368768692, 0.11283446103334427], "decide_seconds": 0.16638362300000153, "probabilities_f64_minus_runtime_max_abs": 3.358997158198207e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.02075878066761299, "darwin": 0.008129255312386804, "sagan": 0.09303440043158626, "ibn_sina": 0.0266548019966612, "einstein": 0.07245526390869803, "socrates": 0.7789674976830546}, "confidence": 0.5413080896477253, "choice": "socrates"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.64182100148173, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "9bca8f1e371997ae4d4ee49bce13b6556ba34f80e746c6c72e2117d9fc507c3a", "compiled_ids_sha256": "25392c527764f88ca74e3177d6be783bb10f5f379082b62d12bcb17e0d187c0e", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 14.5625, 17.0, 15.75, 16.75, 19.125], "runtime_probabilities": [0.02075878158211708, 0.008129255846142769, 0.09303440898656845, 0.02665480226278305, 0.0724552646279335, 0.778967559337616], "decide_seconds": 0.16660703199886484, "probabilities_f64_minus_runtime_max_abs": 6.165456134521463e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.015797622299323296, "einstein": 0.01230320061737927, "darwin": 0.017901051270612194, "sagan": 0.1498836672185462, "hypatia": 0.042942389629109914, "socrates": 0.7611720689650291}, "confidence": 0.5429009237237479, "choice": "socrates"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.5277520013624, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "feef5a6a106e90235b1f563bc78fdb009e737e32d309db116fc776e8272ecd4c", "compiled_ids_sha256": "5132c6740c56f42ab95ac4f4a2d9e7a148e2b54be1bcb3e4fda88f8218fd090a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 14.875, 15.25, 17.375, 16.125, 19.0], "runtime_probabilities": [0.015797622501850128, 0.012303200550377369, 0.01790105178952217, 0.14988365769386292, 0.04294239357113838, 0.7611720561981201], "decide_seconds": 0.16648863199952757, "probabilities_f64_minus_runtime_max_abs": 1.2766908974448654e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.042000222611868684, "sagan": 0.08891447196706205, "socrates": 0.579795188876874, "ibn_sina": 0.0539293533401826, "darwin": 0.16611410280968433, "hypatia": 0.0692466603943284}, "confidence": 0.27171051797493195, "choice": "socrates"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.9253790031653, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "61380b6f56927125bd98ea094e716be7e591d0ff8fbe658560560b004d22de75", "compiled_ids_sha256": "ef14bd6259424374d629ed9e6794321d893025b526a5341852bf84399f21a58f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 17.0, 18.875, 16.5, 17.625, 16.75], "runtime_probabilities": [0.04200022667646408, 0.08891447633504868, 0.5797951817512512, 0.05392935127019882, 0.16611410677433014, 0.06924666464328766], "decide_seconds": 0.16692402999979095, "probabilities_f64_minus_runtime_max_abs": 7.125622758508143e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.06499649493312154, "ibn_sina": 0.011294697234985406, "hypatia": 0.05061932115081165, "einstein": 0.0573592054573351, "sagan": 0.791819406989647, "darwin": 0.02391087423409923}, "confidence": 0.5438144703153123, "choice": "sagan"}}, "usage": {"input_tokens": 343, "output_tokens": 1}, "latency_ms": 170.92998899897793, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 343, "compiled_prompt_sha256": "e59ecb5de50a329c375c493f61934017ab12c03525749ada6d1185884cb8f983", "compiled_ids_sha256": "0c39e6a026de2ad02557d14e5fb1b266a0349c66b025dd19458e82c756828afa", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 14.875, 16.375, 16.5, 19.125, 15.625], "runtime_probabilities": [0.06499648839235306, 0.011294696480035782, 0.05061931535601616, 0.057359203696250916, 0.7918193936347961, 0.023910872638225555], "decide_seconds": 0.16670060099932016, "probabilities_f64_minus_runtime_max_abs": 1.3354850891822423e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:31Z"} {"id": "c-socrates-01-dessert-02", "uid": "c094", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "prompt_sha256_by_order": ["d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6"], "request_sha256": "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "compiled_prompt_sha256_by_order": ["e0d5ffa591269263ed024ca529109dfb26c248d9d18c4a8d6e57f2ba9168b233", "4622674a2ae1102168d7282f9df1a3846c761ba064fef823a6e8d4ebe07ff5f0", "3bd5e7b473e603250aa7e4fad4fa980d5d6ff694a3f53fb5a6faf31bab59e8ad", "8a354122519a3aca753b0519c384a81734212a386aeafce73bd4b02d6bdc04a1", "3c5bdd5f794315afa0e3cd15c25303a46f20d73dfb7a1d058fd2fe4ed4fb7347", "cc2449b92f57f5c6ff39f6eff7855ff197f476dc98099ffbb66a8df50ae2073b"], "compiled_tokens_by_order": [374, 374, 374, 374, 374, 374], "choice": "socrates", "probs": {"darwin": 0.0013152110675347275, "einstein": 0.001090348271539418, "hypatia": 0.002457137760091872, "ibn_sina": 0.0016887644390236078, "sagan": 0.002168416462504728, "socrates": 0.9912801219993056}, "confidence": 0.9644541349943043, "seconds": 0.1718947544989836, "seconds_all_orders": 1.0313685269939015, "seconds_by_order": [0.17118472699803533, 0.17100041800222243, 0.17099439799858374, 0.17106174799846485, 0.17175460199723602, 0.1753726339993591], "argmax_stable": true, "spread": {"darwin": 0.0010967402357164263, "einstein": 0.001242769501518215, "hypatia": 0.005720356317828296, "ibn_sina": 0.0019248390581698498, "sagan": 0.001321933484723786, "socrates": 0.005682355795179839}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.01, "ibn_sina": 0.0, "sagan": 0.0, "socrates": 0.0}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0013152110675347275, "einstein": 0.001090348271539418, "hypatia": 0.002457137760091872, "ibn_sina": 0.0016887644390236078, "sagan": 0.002168416462504728, "socrates": 0.9912801219993056}, {"ibn_sina": 0.0026073684431216554, "hypatia": 0.00487120555339838, "sagan": 0.0017920163768154226, "socrates": 0.9881568629947827, "darwin": 0.0014856337566361256, "einstein": 0.0010869128752457012}, {"hypatia": 0.00517776387043977, "darwin": 0.0016809738159371899, "sagan": 0.0015791287609388117, "ibn_sina": 0.002950200923888889, "einstein": 0.001904792879175067, "socrates": 0.9867071397496203}, {"ibn_sina": 0.0013110584832651665, "einstein": 0.000959190876819875, "darwin": 0.0007017590366244536, "sagan": 0.0008464829777809419, "hypatia": 0.00803120688315972, "socrates": 0.9881503017423497}, {"einstein": 0.0014919972623012945, "sagan": 0.0010254337223651617, "socrates": 0.9923894955448002, "ibn_sina": 0.0016906543897566519, "darwin": 0.0010915685154452293, "hypatia": 0.0023108505653314233}, {"socrates": 0.9923199544528294, "ibn_sina": 0.0010253618657190392, "hypatia": 0.004316934699130323, "einstein": 0.0006620233776568518, "sagan": 0.0010914920244435955, "darwin": 0.0005842335802207636}], "prompt_tokens": 374, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0013152110675347275, "einstein": 0.001090348271539418, "hypatia": 0.002457137760091872, "ibn_sina": 0.0016887644390236078, "sagan": 0.002168416462504728, "socrates": 0.9912801219993056}, "confidence": 0.9644541349943043, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 171.15643700162764, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "e0d5ffa591269263ed024ca529109dfb26c248d9d18c4a8d6e57f2ba9168b233", "compiled_ids_sha256": "42ef38f2cbdef5189c3d759a7044b4c8a748d18138cc275d8d40d3f3af6771d4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.0, 12.8125, 13.625, 13.25, 13.5, 19.625], "runtime_probabilities": [0.0013152110623195767, 0.001090348232537508, 0.002457137918099761, 0.001688764663413167, 0.0021684165112674236, 0.9912801384925842], "decide_seconds": 0.16679143099827343, "probabilities_f64_minus_runtime_max_abs": 1.649327863173511e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0026073684431216554, "hypatia": 0.00487120555339838, "sagan": 0.0017920163768154226, "socrates": 0.9881568629947827, "darwin": 0.0014856337566361256, "einstein": 0.0010869128752457012}, "confidence": 0.9544321247252576, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 170.97593799917377, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "4622674a2ae1102168d7282f9df1a3846c761ba064fef823a6e8d4ebe07ff5f0", "compiled_ids_sha256": "392b0391b76bd9c054c06c3726dfba5edfa6caf992952cb90af705f0ebe4c781", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.9375, 14.5625, 13.5625, 19.875, 13.375, 13.0625], "runtime_probabilities": [0.0026073686312884092, 0.004871205426752567, 0.0017920163227245212, 0.9881568551063538, 0.0014856337802484632, 0.0010869128163903952], "decide_seconds": 0.1667098320031073, "probabilities_f64_minus_runtime_max_abs": 7.888428910085565e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.00517776387043977, "darwin": 0.0016809738159371899, "sagan": 0.0015791287609388117, "ibn_sina": 0.002950200923888889, "einstein": 0.001904792879175067, "socrates": 0.9867071397496203}, "confidence": 0.9494909272862266, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 170.9701479994692, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "3bd5e7b473e603250aa7e4fad4fa980d5d6ff694a3f53fb5a6faf31bab59e8ad", "compiled_ids_sha256": "bb3ae6e0798e18997aacf12d2b357fb0ce288f0a4f72d30e266ecea61c5ee1ba", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 13.375, 13.3125, 13.9375, 13.5, 19.75], "runtime_probabilities": [0.005177764222025871, 0.0016809741500765085, 0.0015791290206834674, 0.002950201276689768, 0.0019047930836677551, 0.9867072105407715], "decide_seconds": 0.1667195609988994, "probabilities_f64_minus_runtime_max_abs": 7.079115116326307e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0013110584832651665, "einstein": 0.000959190876819875, "darwin": 0.0007017590366244536, "sagan": 0.0008464829777809419, "hypatia": 0.00803120688315972, "socrates": 0.9881503017423497}, "confidence": 0.9570384093469259, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 171.03741800019634, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "8a354122519a3aca753b0519c384a81734212a386aeafce73bd4b02d6bdc04a1", "compiled_ids_sha256": "92b126d4b011e122a037ec61f29c4d0697dd7f29734e6c00331be0ff987b79f4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.125, 12.8125, 12.5, 12.6875, 14.9375, 19.75], "runtime_probabilities": [0.0013110584113746881, 0.0009591907728463411, 0.00070175901055336, 0.0008464829297736287, 0.008031207136809826, 0.9881502389907837], "decide_seconds": 0.16679426999689895, "probabilities_f64_minus_runtime_max_abs": 6.27515660545086e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0014919972623012945, "sagan": 0.0010254337223651617, "socrates": 0.9923894955448002, "ibn_sina": 0.0016906543897566519, "darwin": 0.0010915685154452293, "hypatia": 0.0023108505653314233}, "confidence": 0.9684046793456855, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 171.72988200036343, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "3c5bdd5f794315afa0e3cd15c25303a46f20d73dfb7a1d058fd2fe4ed4fb7347", "compiled_ids_sha256": "c06290d6f78e341cb8c609a0f08a06e55061bd9f3aa5755f6d95f0a90134e3f5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.375, 13.0, 19.875, 13.5, 13.0625, 13.8125], "runtime_probabilities": [0.001491997274570167, 0.001025433768518269, 0.9923895001411438, 0.0016906545497477055, 0.001091568497940898, 0.0023108504246920347], "decide_seconds": 0.16748630499932915, "probabilities_f64_minus_runtime_max_abs": 4.596343639029499e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9923199544528294, "ibn_sina": 0.0010253618657190392, "hypatia": 0.004316934699130323, "einstein": 0.0006620233776568518, "sagan": 0.0010914920244435955, "darwin": 0.0005842335802207636}, "confidence": 0.9693851447230044, "choice": "socrates"}}, "usage": {"input_tokens": 374, "output_tokens": 1}, "latency_ms": 175.34547499963082, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 374, "compiled_prompt_sha256": "cc2449b92f57f5c6ff39f6eff7855ff197f476dc98099ffbb66a8df50ae2073b", "compiled_ids_sha256": "6bcba28bb55de8670937e51ef0c42644a098379968dfb4426183c8d33b7c67f9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 12.875, 14.3125, 12.4375, 12.9375, 12.3125], "runtime_probabilities": [0.9923200011253357, 0.0010253619402647018, 0.0043169348500669, 0.0006620234344154596, 0.0010914920130744576, 0.0005842336104251444], "decide_seconds": 0.1705840709983022, "probabilities_f64_minus_runtime_max_abs": 4.667250630419062e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:32Z"} {"id": "c-socrates-01-dessert-02", "uid": "c095", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "prompt_sha256_by_order": ["1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191"], "request_sha256": "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "compiled_prompt_sha256_by_order": ["3b30247295b5c3dec8716911428c2b205adc94f0d63fb7499beab302bd9b5b24", "47d94f6baeffe2d38319a82c77c681fea06cfc8fccfd89082f20513cf2c5b6ef", "679e4cc68c56d8be4d869df90871cd1c3221655bdd6e47c48bf51899a6f41c3a", "36007f153866a07fe2d63f094e5a85a563b8b67238b89be7a2bf35504006f7e5", "b1aee04d9d360303f87dece0913ca09699b7b9c0711e1c22e65b5bca4644f986", "46362ab9da0c35611866d89b74624466b224976f9884333b34bd20536e18016d"], "compiled_tokens_by_order": [302, 302, 302, 302, 302, 302], "choice": "hypatia", "probs": {"darwin": 0.012909095301011576, "einstein": 0.005728376315404194, "hypatia": 0.5843102694324854, "ibn_sina": 0.022656167825184672, "sagan": 0.019993997930162242, "socrates": 0.35440209319575194}, "confidence": 0.48020581373055304, "seconds": 0.16540236649962026, "seconds_all_orders": 0.9924141989977215, "seconds_by_order": [0.16612820600130362, 0.16519493300074828, 0.16521709299922804, 0.16512675400008447, 0.16511405399796786, 0.16563315899838926], "argmax_stable": true, "spread": {"darwin": 0.014988619690753968, "einstein": 0.00800908587906287, "hypatia": 0.3194432555509964, "ibn_sina": 0.10532436408793715, "sagan": 0.04246951158445402, "socrates": 0.2726912335448464}, "spread_2dp": {"darwin": 0.02, "einstein": 0.01, "hypatia": 0.31999999999999995, "ibn_sina": 0.1, "sagan": 0.04, "socrates": 0.26999999999999996}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.012909095301011576, "einstein": 0.005728376315404194, "hypatia": 0.5843102694324854, "ibn_sina": 0.022656167825184672, "sagan": 0.019993997930162242, "socrates": 0.35440209319575194}, {"ibn_sina": 0.11449206846251296, "hypatia": 0.581438700086151, "sagan": 0.028948128312911138, "socrates": 0.2423797108371594, "darwin": 0.019895738230529217, "einstein": 0.012845654070736266}, {"hypatia": 0.8784745295525197, "darwin": 0.004907118539775249, "sagan": 0.017127526636734974, "ibn_sina": 0.009167704374575819, "einstein": 0.008612261245488771, "socrates": 0.08171085965090553}, {"ibn_sina": 0.11007978465517478, "einstein": 0.004836568191673396, "darwin": 0.005148500040214702, "sagan": 0.0216759948025674, "hypatia": 0.5590312740015233, "socrates": 0.29922787830884645}, {"einstein": 0.011501437855406365, "sagan": 0.005103732110401278, "socrates": 0.1025111997495234, "ibn_sina": 0.01303283651389796, "darwin": 0.009535026883074213, "hypatia": 0.8583157668876967}, {"socrates": 0.16604693609395413, "ibn_sina": 0.01644087309103285, "hypatia": 0.7441707386553974, "einstein": 0.00826699015958403, "sagan": 0.0475732436948553, "darwin": 0.01750121830517621}], "prompt_tokens": 302, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.012909095301011576, "einstein": 0.005728376315404194, "hypatia": 0.5843102694324854, "ibn_sina": 0.022656167825184672, "sagan": 0.019993997930162242, "socrates": 0.35440209319575194}, "confidence": 0.48020581373055304, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 166.09979599888902, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "3b30247295b5c3dec8716911428c2b205adc94f0d63fb7499beab302bd9b5b24", "compiled_ids_sha256": "0da5ad6c304119d9f5f455fa27621e3de8086264734e81f84f183030922025e2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 14.875, 19.5, 16.25, 16.125, 19.0], "runtime_probabilities": [0.01290909480303526, 0.005728376563638449, 0.5843102335929871, 0.022656166926026344, 0.019993996247649193, 0.354402095079422], "decide_seconds": 0.16211402599947178, "probabilities_f64_minus_runtime_max_abs": 3.5839498369050204e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.11449206846251296, "hypatia": 0.581438700086151, "sagan": 0.028948128312911138, "socrates": 0.2423797108371594, "darwin": 0.019895738230529217, "einstein": 0.012845654070736266}, "confidence": 0.3618852502590155, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.1713529972767, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "47d94f6baeffe2d38319a82c77c681fea06cfc8fccfd89082f20513cf2c5b6ef", "compiled_ids_sha256": "bc469acfacc2210432ab12540c44a63f4f17495da4ce3a8ba0122b320841a6cd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.625, 19.25, 16.25, 18.375, 15.875, 15.4375], "runtime_probabilities": [0.11449206620454788, 0.5814387202262878, 0.02894813008606434, 0.24237972497940063, 0.01989573985338211, 0.012845654971897602], "decide_seconds": 0.16143774199736072, "probabilities_f64_minus_runtime_max_abs": 2.014013689155547e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.8784745295525197, "darwin": 0.004907118539775249, "sagan": 0.017127526636734974, "ibn_sina": 0.009167704374575819, "einstein": 0.008612261245488771, "socrates": 0.08171085965090553}, "confidence": 0.7219569154570524, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.19462300129817, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "679e4cc68c56d8be4d869df90871cd1c3221655bdd6e47c48bf51899a6f41c3a", "compiled_ids_sha256": "be184a868b536337afe9e9b09ac9a7e7954a242d364f2ce1cac676ffa9a7541d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.875, 14.6875, 15.9375, 15.3125, 15.25, 17.5], "runtime_probabilities": [0.8784745335578918, 0.004907119087874889, 0.01712752878665924, 0.009167704731225967, 0.008612262085080147, 0.08171086013317108], "decide_seconds": 0.16154747100154054, "probabilities_f64_minus_runtime_max_abs": 4.005372145066133e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.11007978465517478, "einstein": 0.004836568191673396, "darwin": 0.005148500040214702, "sagan": 0.0216759948025674, "hypatia": 0.5590312740015233, "socrates": 0.29922787830884645}, "confidence": 0.4056115068389531, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.10447399923578, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "36007f153866a07fe2d63f094e5a85a563b8b67238b89be7a2bf35504006f7e5", "compiled_ids_sha256": "314d191f7ccedde6050290f5578f4ecb71047fbf321ae845a84c026df4b90fbd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.875, 14.75, 14.8125, 16.25, 19.5, 18.875], "runtime_probabilities": [0.11007978767156601, 0.004836568608880043, 0.005148500204086304, 0.021675996482372284, 0.5590313076972961, 0.29922792315483093], "decide_seconds": 0.16150331099925097, "probabilities_f64_minus_runtime_max_abs": 4.484598448151189e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.011501437855406365, "sagan": 0.005103732110401278, "socrates": 0.1025111997495234, "ibn_sina": 0.01303283651389796, "darwin": 0.009535026883074213, "hypatia": 0.8583157668876967}, "confidence": 0.6964666680852983, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.0907939983881, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "b1aee04d9d360303f87dece0913ca09699b7b9c0711e1c22e65b5bca4644f986", "compiled_ids_sha256": "c6a725bf78557eba707736aeda915f5ff3cc72245ebae27367c3f2a38f0817bc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.75, 17.75, 15.6875, 15.375, 19.875], "runtime_probabilities": [0.011501438915729523, 0.005103731993585825, 0.1025112047791481, 0.013032836839556694, 0.009535027667880058, 0.8583158254623413], "decide_seconds": 0.16153349099840852, "probabilities_f64_minus_runtime_max_abs": 5.857464457914574e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.16604693609395413, "ibn_sina": 0.01644087309103285, "hypatia": 0.7441707386553974, "einstein": 0.00826699015958403, "sagan": 0.0475732436948553, "darwin": 0.01750121830517621}, "confidence": 0.5306884239750231, "choice": "hypatia"}}, "usage": {"input_tokens": 302, "output_tokens": 1}, "latency_ms": 165.60994899919024, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 302, "compiled_prompt_sha256": "46362ab9da0c35611866d89b74624466b224976f9884333b34bd20536e18016d", "compiled_ids_sha256": "9e74852718ba36a208eb2d3c877352b48ff0e0ecb913eff96b21ccc5d7664118", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.25, 15.9375, 19.75, 15.25, 17.0, 16.0], "runtime_probabilities": [0.16604693233966827, 0.016440872102975845, 0.7441707253456116, 0.008266991004347801, 0.04757324233651161, 0.017501218244433403], "decide_seconds": 0.16198035799970967, "probabilities_f64_minus_runtime_max_abs": 1.3309785829029863e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:33Z"} {"id": "c-socrates-01-main-01", "uid": "c096", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "prompt_sha256_by_order": ["d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79"], "request_sha256": "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "compiled_prompt_sha256_by_order": ["1370e6b93e0997fcf22f6ff3e6473a4bccf01f168816a3e2da10413c14e6b44c", "8ac4d63f23757aa1e7bbb97c8d49528f90a81b241b6099a9916007075b7c9f98", "18b2c5ba46c76521f3fa6d8c28a5cbe620cab72d2828a21c063943ed1890e141", "38e30e3c151e24732f60c596fac287a3385502ec8dc67fdbd09c5c1d4df919ac", "d703416e934056107402f89bd2087c2fd95404a6eb696db67485e9b3c16edf8c", "370f936f1ef7b1e357129cfe9f3030c4305fefb65d4e8856c5652becf441ae71"], "compiled_tokens_by_order": [323, 323, 323, 323, 323, 323], "choice": "socrates", "probs": {"darwin": 0.061328951546661084, "einstein": 0.0477630354894879, "hypatia": 0.028969745425317016, "ibn_sina": 0.01368433878034178, "sagan": 0.10111434692471766, "socrates": 0.7471395818334746}, "confidence": 0.48246864673598644, "seconds": 0.1706958771671149, "seconds_all_orders": 1.0241752630026895, "seconds_by_order": [0.17091017900020233, 0.17057822199785733, 0.1705337210005382, 0.17043290299989167, 0.1706766500028607, 0.17104358800133923], "argmax_stable": true, "spread": {"darwin": 0.12576474225246734, "einstein": 0.051221732934454214, "hypatia": 0.03464292483022974, "ibn_sina": 0.022380271753424552, "sagan": 0.08596696882031635, "socrates": 0.1781892899802323}, "spread_2dp": {"darwin": 0.13, "einstein": 0.05, "hypatia": 0.04, "ibn_sina": 0.019999999999999997, "sagan": 0.08, "socrates": 0.17999999999999994}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.061328951546661084, "einstein": 0.0477630354894879, "hypatia": 0.028969745425317016, "ibn_sina": 0.01368433878034178, "sagan": 0.10111434692471766, "socrates": 0.7471395818334746}, {"ibn_sina": 0.016124300059268907, "hypatia": 0.01336751425941188, "sagan": 0.015147378104401314, "socrates": 0.827018822352867, "darwin": 0.08716714257405841, "einstein": 0.041174842649992514}, {"hypatia": 0.03944615393469742, "darwin": 0.12150270797466929, "sagan": 0.017504124701566183, "ibn_sina": 0.01061678830293314, "einstein": 0.018633043753024433, "socrates": 0.7922971813331096}, {"ibn_sina": 0.016395972456271356, "einstein": 0.022410637235672215, "darwin": 0.060918427961915426, "sagan": 0.017453421828270706, "hypatia": 0.04186858241902465, "socrates": 0.8409529580988456}, {"einstein": 0.06985477668747865, "sagan": 0.01880115808672739, "socrates": 0.6627636681186133, "ibn_sina": 0.03299706005635769, "darwin": 0.1675728979611813, "hypatia": 0.048010439089641616}, {"socrates": 0.8397392551777536, "ibn_sina": 0.011252512456101548, "hypatia": 0.014448511995229793, "einstein": 0.023821569056501308, "sagan": 0.0689299956056997, "darwin": 0.04180815570871397}], "prompt_tokens": 323, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.061328951546661084, "einstein": 0.0477630354894879, "hypatia": 0.028969745425317016, "ibn_sina": 0.01368433878034178, "sagan": 0.10111434692471766, "socrates": 0.7471395818334746}, "confidence": 0.48246864673598644, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.87934899973334, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "1370e6b93e0997fcf22f6ff3e6473a4bccf01f168816a3e2da10413c14e6b44c", "compiled_ids_sha256": "a31f962774edf9fb5117efe523b6eb2f11ca98ce116e6d7c52e3734ea7a26aa9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.625, 16.375, 15.875, 15.125, 17.125, 19.125], "runtime_probabilities": [0.0613289512693882, 0.04776303097605705, 0.028969744220376015, 0.013684338890016079, 0.10111434012651443, 0.747139573097229], "decide_seconds": 0.1668650899991917, "probabilities_f64_minus_runtime_max_abs": 8.736245615814653e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.016124300059268907, "hypatia": 0.01336751425941188, "sagan": 0.015147378104401314, "socrates": 0.827018822352867, "darwin": 0.08716714257405841, "einstein": 0.041174842649992514}, "confidence": 0.615574313845286, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.55453200009651, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "8ac4d63f23757aa1e7bbb97c8d49528f90a81b241b6099a9916007075b7c9f98", "compiled_ids_sha256": "034dbc5dacc23f547fde305251460e794274f26f1bb333d000f71eac1d884f6b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.4375, 15.25, 15.375, 19.375, 17.125, 16.375], "runtime_probabilities": [0.016124302521348, 0.013367515057325363, 0.015147379599511623, 0.8270188570022583, 0.08716714382171631, 0.04117484390735626], "decide_seconds": 0.16677535999770043, "probabilities_f64_minus_runtime_max_abs": 3.464939135611189e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.03944615393469742, "darwin": 0.12150270797466929, "sagan": 0.017504124701566183, "ibn_sina": 0.01061678830293314, "einstein": 0.018633043753024433, "socrates": 0.7922971813331096}, "confidence": 0.5750723967251453, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.510572999774, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "18b2c5ba46c76521f3fa6d8c28a5cbe620cab72d2828a21c063943ed1890e141", "compiled_ids_sha256": "d1c9fe1dfb59f3aaaf5898ff6ffad171a178715b36299ccdd6a18344b5d63269", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 17.375, 15.4375, 14.9375, 15.5, 19.25], "runtime_probabilities": [0.0394461527466774, 0.12150271236896515, 0.017504123970866203, 0.010616788640618324, 0.0186330433934927, 0.7922971844673157], "decide_seconds": 0.16664267200030736, "probabilities_f64_minus_runtime_max_abs": 4.394295857745867e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.016395972456271356, "einstein": 0.022410637235672215, "darwin": 0.060918427961915426, "sagan": 0.017453421828270706, "hypatia": 0.04186858241902465, "socrates": 0.8409529580988456}, "confidence": 0.6248570750939244, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.41019299722393, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "38e30e3c151e24732f60c596fac287a3385502ec8dc67fdbd09c5c1d4df919ac", "compiled_ids_sha256": "a160c69a210a975c030748f373b8c350d4f1bdb9fe6b77b7bf562a76269a02c2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.875, 16.875, 15.625, 16.5, 19.5], "runtime_probabilities": [0.016395973041653633, 0.022410636767745018, 0.060918428003787994, 0.017453420907258987, 0.041868578642606735, 0.8409529328346252], "decide_seconds": 0.16661161199954222, "probabilities_f64_minus_runtime_max_abs": 2.5264220382403835e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.06985477668747865, "sagan": 0.01880115808672739, "socrates": 0.6627636681186133, "ibn_sina": 0.03299706005635769, "darwin": 0.1675728979611813, "hypatia": 0.048010439089641616}, "confidence": 0.3911456754502771, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 170.65385100067942, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "d703416e934056107402f89bd2087c2fd95404a6eb696db67485e9b3c16edf8c", "compiled_ids_sha256": "85633e66aba1a86d4e40aa43553b5b4e69a92fbd21c9aef9c0906a955509e903", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.125, 15.8125, 19.375, 16.375, 18.0, 16.75], "runtime_probabilities": [0.06985477358102798, 0.018801160156726837, 0.6627637147903442, 0.03299706056714058, 0.16757290065288544, 0.0480104424059391], "decide_seconds": 0.16686784999910742, "probabilities_f64_minus_runtime_max_abs": 4.667173092443022e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.8397392551777536, "ibn_sina": 0.011252512456101548, "hypatia": 0.014448511995229793, "einstein": 0.023821569056501308, "sagan": 0.0689299956056997, "darwin": 0.04180815570871397}, "confidence": 0.629134829144903, "choice": "socrates"}}, "usage": {"input_tokens": 323, "output_tokens": 1}, "latency_ms": 171.01958799685235, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 323, "compiled_prompt_sha256": "370f936f1ef7b1e357129cfe9f3030c4305fefb65d4e8856c5652becf441ae71", "compiled_ids_sha256": "3828151f5a960fdb5cfad4afb2580cf6b214834f845ce0ea9c9fc791a4760a20", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 14.9375, 15.1875, 15.6875, 16.75, 16.25], "runtime_probabilities": [0.8397392630577087, 0.011252512224018574, 0.01444851141422987, 0.023821569979190826, 0.0689299926161766, 0.04180815443396568], "decide_seconds": 0.1670749290024105, "probabilities_f64_minus_runtime_max_abs": 7.879955132850114e-09}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:34Z"} {"id": "c-socrates-01-main-02", "uid": "c097", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "prompt_sha256_by_order": ["b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d"], "request_sha256": "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "compiled_prompt_sha256_by_order": ["7a7cb526fa35845664c998ea567eaa951e2590b39bd2a7c881e769dec476a428", "6b3394ef3d36b8a88c04b2cd2d19c899848b44b4390af0213d9524d139bfd7d9", "c95bdf299bf010e460bb20a55fe1c0478779868b109f9c5024421795c18a8ba0", "300d217cf43c51333e1e6f2180bc908f71205ee809d827928911f8a4767deb6b", "8b631b1e7ddb293360b4bb46c71d2ee03870dda007d5250683d2791eaf0e7a95", "52541a1b16e42ec70681506c471705cfc38ab8ff7d555574f41a36b30d55f23f"], "compiled_tokens_by_order": [395, 395, 395, 395, 395, 395], "choice": "socrates", "probs": {"darwin": 0.04000983549462163, "einstein": 0.013827029159067236, "hypatia": 0.12323896941297224, "ibn_sina": 0.09597860588373569, "sagan": 0.01775425687752486, "socrates": 0.7091913031720783}, "confidence": 0.4495933078619099, "seconds": 0.17693717633422543, "seconds_all_orders": 1.0616230580053525, "seconds_by_order": [0.17743383899869514, 0.176871954001399, 0.17678504400100792, 0.17689455300205736, 0.17680885400113766, 0.17682881400105543], "argmax_stable": false, "spread": {"darwin": 0.03968896678818072, "einstein": 0.024489520420366152, "hypatia": 0.21382390664600487, "ibn_sina": 0.34924035237201534, "sagan": 0.02304423467291808, "socrates": 0.4479551073548211}, "spread_2dp": {"darwin": 0.04, "einstein": 0.03, "hypatia": 0.22000000000000003, "ibn_sina": 0.35000000000000003, "sagan": 0.019999999999999997, "socrates": 0.43999999999999995}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "ibn_sina"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.04000983549462163, "einstein": 0.013827029159067236, "hypatia": 0.12323896941297224, "ibn_sina": 0.09597860588373569, "sagan": 0.01775425687752486, "socrates": 0.7091913031720783}, {"ibn_sina": 0.06637161557391683, "hypatia": 0.3370628760589771, "sagan": 0.013069346008668735, "socrates": 0.49042359085233056, "darwin": 0.07520889351512992, "einstein": 0.017863677990976747}, {"hypatia": 0.12572617515453247, "darwin": 0.046252075056472275, "sagan": 0.009107573581040074, "ibn_sina": 0.05938883994705951, "einstein": 0.036021152272658, "socrates": 0.7235041839882377}, {"ibn_sina": 0.05168118843557867, "einstein": 0.011531631852291849, "darwin": 0.0355199267269492, "sagan": 0.007445381112741176, "hypatia": 0.1803850720893804, "socrates": 0.7134367997830587}, {"einstein": 0.018476447697410314, "sagan": 0.005998424332981758, "socrates": 0.5747854401354809, "ibn_sina": 0.06864833123187337, "darwin": 0.060581939679729586, "hypatia": 0.271509416922524}, {"socrates": 0.2755490766334166, "ibn_sina": 0.400921540807594, "hypatia": 0.21459783665670742, "einstein": 0.025630056615527214, "sagan": 0.029042659005899838, "darwin": 0.05425883028085491}], "prompt_tokens": 395, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.04000983549462163, "einstein": 0.013827029159067236, "hypatia": 0.12323896941297224, "ibn_sina": 0.09597860588373569, "sagan": 0.01775425687752486, "socrates": 0.7091913031720783}, "confidence": 0.4495933078619099, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 177.40572900220286, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "7a7cb526fa35845664c998ea567eaa951e2590b39bd2a7c881e769dec476a428", "compiled_ids_sha256": "fba0ef685096385bfc59e52dd6d585816ee5d2066e5bbea9c00a20c6d1da8968", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.5, 15.4375, 17.625, 17.375, 15.6875, 19.375], "runtime_probabilities": [0.040009837597608566, 0.01382702961564064, 0.1232389509677887, 0.09597860276699066, 0.01775425672531128, 0.7091912627220154], "decide_seconds": 0.17265129500083276, "probabilities_f64_minus_runtime_max_abs": 4.045006296582443e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.06637161557391683, "hypatia": 0.3370628760589771, "sagan": 0.013069346008668735, "socrates": 0.49042359085233056, "darwin": 0.07520889351512992, "einstein": 0.017863677990976747}, "confidence": 0.319554823754976, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.84637400088832, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "6b3394ef3d36b8a88c04b2cd2d19c899848b44b4390af0213d9524d139bfd7d9", "compiled_ids_sha256": "20a28d91a25d65cc9539fac61a4984b1761b0bf02bb9bd37e903487ffc0140f2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.0, 18.625, 15.375, 19.0, 17.125, 15.6875], "runtime_probabilities": [0.0663716122508049, 0.33706286549568176, 0.013069345615804195, 0.4904235899448395, 0.07520889490842819, 0.017863677814602852], "decide_seconds": 0.17240884700004244, "probabilities_f64_minus_runtime_max_abs": 1.056329534954159e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.12572617515453247, "darwin": 0.046252075056472275, "sagan": 0.009107573581040074, "ibn_sina": 0.05938883994705951, "einstein": 0.036021152272658, "socrates": 0.7235041839882377}, "confidence": 0.4601707591982809, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.75996399702854, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "c95bdf299bf010e460bb20a55fe1c0478779868b109f9c5024421795c18a8ba0", "compiled_ids_sha256": "6315a386fe3572227e2adc7b826a6a1623c6ea83be4e824cd61b299325b2b407", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.75, 16.75, 15.125, 17.0, 16.5, 19.5], "runtime_probabilities": [0.12572616338729858, 0.04625207185745239, 0.009107572957873344, 0.05938883498311043, 0.0360211506485939, 0.7235041260719299], "decide_seconds": 0.17227811900011147, "probabilities_f64_minus_runtime_max_abs": 5.7916307749650286e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.05168118843557867, "einstein": 0.011531631852291849, "darwin": 0.0355199267269492, "sagan": 0.007445381112741176, "hypatia": 0.1803850720893804, "socrates": 0.7134367997830587}, "confidence": 0.49242554817680906, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.8701630026044, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "300d217cf43c51333e1e6f2180bc908f71205ee809d827928911f8a4767deb6b", "compiled_ids_sha256": "2a36b86eacc3196d5219396270e81598da67bf34fd583de0dcb05fd710af56ac", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 15.375, 16.5, 14.9375, 18.125, 19.5], "runtime_probabilities": [0.051681190729141235, 0.011531632393598557, 0.035519927740097046, 0.007445381488651037, 0.18038508296012878, 0.7134368419647217], "decide_seconds": 0.17245208700114745, "probabilities_f64_minus_runtime_max_abs": 4.2181662940343756e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.018476447697410314, "sagan": 0.005998424332981758, "socrates": 0.5747854401354809, "ibn_sina": 0.06864833123187337, "darwin": 0.060581939679729586, "hypatia": 0.271509416922524}, "confidence": 0.369079273393058, "choice": "socrates"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.78401399825816, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "8b631b1e7ddb293360b4bb46c71d2ee03870dda007d5250683d2791eaf0e7a95", "compiled_ids_sha256": "72191dec5f384150c83643ceaae301013348ed8050ea27e46a363d8d127ee808", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.9375, 14.8125, 19.375, 17.25, 17.125, 18.625], "runtime_probabilities": [0.018476447090506554, 0.005998424254357815, 0.5747854113578796, 0.0686483308672905, 0.060581933706998825, 0.27150940895080566], "decide_seconds": 0.17237097800170886, "probabilities_f64_minus_runtime_max_abs": 2.877760130193252e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.2755490766334166, "ibn_sina": 0.400921540807594, "hypatia": 0.21459783665670742, "einstein": 0.025630056615527214, "sagan": 0.029042659005899838, "darwin": 0.05425883028085491}, "confidence": 0.21491617507561078, "choice": "ibn_sina"}}, "usage": {"input_tokens": 395, "output_tokens": 1}, "latency_ms": 176.803813999868, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 395, "compiled_prompt_sha256": "52541a1b16e42ec70681506c471705cfc38ab8ff7d555574f41a36b30d55f23f", "compiled_ids_sha256": "5e60fa068aef24d30a8d982a4dfeeb7ef924cb25b06ba0a127552ec31504ea4a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [18.625, 19.0, 18.375, 16.25, 16.375, 17.0], "runtime_probabilities": [0.2755490839481354, 0.4009215235710144, 0.21459785103797913, 0.025630054995417595, 0.02904265932738781, 0.05425883084535599], "decide_seconds": 0.17234087800170528, "probabilities_f64_minus_runtime_max_abs": 1.72365796080598e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:35Z"} {"id": "c-socrates-01-main-02", "uid": "c098", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "prompt_sha256_by_order": ["7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53"], "request_sha256": "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "compiled_prompt_sha256_by_order": ["ae667f183d359d40c186509ad74566d963a401022b071a698cabca10ad469305", "49640ee5153760731e8f22301aa321ba4a52e2aee38de52860b386825a269840", "44a75d78e0aaaecb479ef43e098bc75854d559cb678c48f8e2cee707acf08adc", "0144f56f396253c962fe0b30ed9f396a34ce328a66dd5ae3c9ff4ced79fe878b", "e4cf79a383e4270bd75dbad0aced9f5f9b58e7bfd06ab3e39727d6af5a53f92e", "1897c93ff7a27e62ef324efcaada8d4989f2f3b41b3a912beb3f9588c66e9713"], "compiled_tokens_by_order": [297, 297, 297, 297, 297, 297], "choice": "socrates", "probs": {"darwin": 0.020353545021783943, "einstein": 0.01585135680124443, "hypatia": 0.11003022483332527, "ibn_sina": 0.01159710037580007, "sagan": 0.12468057906047764, "socrates": 0.7174871939073686}, "confidence": 0.47689479434562565, "seconds": 0.16528881866724987, "seconds_all_orders": 0.9917329120034992, "seconds_by_order": [0.165876297000068, 0.16509281400067266, 0.16502292400036822, 0.1652104530003271, 0.1652287620017887, 0.16530166200027452], "argmax_stable": false, "spread": {"darwin": 0.02883881806343651, "einstein": 0.12902976167985752, "hypatia": 0.10999926236412128, "ibn_sina": 0.019953692385509484, "sagan": 0.6125118484342006, "socrates": 0.60630210515479}, "spread_2dp": {"darwin": 0.03, "einstein": 0.13, "hypatia": 0.11000000000000001, "ibn_sina": 0.019999999999999997, "sagan": 0.6100000000000001, "socrates": 0.6}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.020353545021783943, "einstein": 0.01585135680124443, "hypatia": 0.11003022483332527, "ibn_sina": 0.01159710037580007, "sagan": 0.12468057906047764, "socrates": 0.7174871939073686}, {"ibn_sina": 0.029097466468089497, "hypatia": 0.0841963109577376, "sagan": 0.10811020326107966, "socrates": 0.6221312649897325, "darwin": 0.017648505532856552, "einstein": 0.13881624879050422}, {"hypatia": 0.063470372331309, "darwin": 0.011740838326483373, "sagan": 0.22153336707406532, "ibn_sina": 0.009143774082580013, "einstein": 0.011740838326483373, "socrates": 0.6823708098590789}, {"ibn_sina": 0.01828356498192823, "einstein": 0.00978648711064669, "darwin": 0.007159946922633981, "sagan": 0.10521465197192885, "hypatia": 0.17346963469543028, "socrates": 0.686085714317432}, {"einstein": 0.023242549566480405, "sagan": 0.05935192955146942, "socrates": 0.7230545233168743, "ibn_sina": 0.015974355128809743, "darwin": 0.03599876498607049, "hypatia": 0.1423778774502957}, {"socrates": 0.11675241816208425, "ibn_sina": 0.02773112999323369, "hypatia": 0.11675241816208425, "einstein": 0.03345012784846391, "sagan": 0.67186377798567, "darwin": 0.03345012784846391}], "prompt_tokens": 297, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.020353545021783943, "einstein": 0.01585135680124443, "hypatia": 0.11003022483332527, "ibn_sina": 0.01159710037580007, "sagan": 0.12468057906047764, "socrates": 0.7174871939073686}, "confidence": 0.47689479434562565, "choice": "socrates"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.85025799940922, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "ae667f183d359d40c186509ad74566d963a401022b071a698cabca10ad469305", "compiled_ids_sha256": "455a8afb9852c21fd5125d01820fb2e27419fa8e5daf8123220881908a4db405", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 15.3125, 17.25, 15.0, 17.375, 19.125], "runtime_probabilities": [0.020353546366095543, 0.015851356089115143, 0.11003023386001587, 0.011597100645303726, 0.12468057125806808, 0.7174872159957886], "decide_seconds": 0.16209597700071754, "probabilities_f64_minus_runtime_max_abs": 2.208841998196931e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.029097466468089497, "hypatia": 0.0841963109577376, "sagan": 0.10811020326107966, "socrates": 0.6221312649897325, "darwin": 0.017648505532856552, "einstein": 0.13881624879050422}, "confidence": 0.3345101087210618, "choice": "socrates"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.06940399995074, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "49640ee5153760731e8f22301aa321ba4a52e2aee38de52860b386825a269840", "compiled_ids_sha256": "70f658f6e03afeae0583917abcec524f436afde8d7cd7e3c0acb04b1fc085bdb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.6875, 16.75, 17.0, 18.75, 15.1875, 17.25], "runtime_probabilities": [0.02909746952354908, 0.0841963142156601, 0.10811019688844681, 0.6221312880516052, 0.01764850690960884, 0.1388162523508072], "decide_seconds": 0.1614623509994999, "probabilities_f64_minus_runtime_max_abs": 2.306187274481175e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.063470372331309, "darwin": 0.011740838326483373, "sagan": 0.22153336707406532, "ibn_sina": 0.009143774082580013, "einstein": 0.011740838326483373, "socrates": 0.6823708098590789}, "confidence": 0.48822591170768515, "choice": "socrates"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.00059400277678, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "44a75d78e0aaaecb479ef43e098bc75854d559cb678c48f8e2cee707acf08adc", "compiled_ids_sha256": "87f83b83d25d1cd65863c60d394606cbe0c6ce75d748516fb7f0bb833675156f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.875, 15.1875, 18.125, 14.9375, 15.1875, 19.25], "runtime_probabilities": [0.06347037106752396, 0.01174083910882473, 0.2215333729982376, 0.009143774397671223, 0.01174083910882473, 0.6823708415031433], "decide_seconds": 0.1613815419987077, "probabilities_f64_minus_runtime_max_abs": 3.164406436884093e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01828356498192823, "einstein": 0.00978648711064669, "darwin": 0.007159946922633981, "sagan": 0.10521465197192885, "hypatia": 0.17346963469543028, "socrates": 0.686085714317432}, "confidence": 0.46807107833761574, "choice": "socrates"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.18826299943612, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "0144f56f396253c962fe0b30ed9f396a34ce328a66dd5ae3c9ff4ced79fe878b", "compiled_ids_sha256": "6d5872db9660d96f58948862d9dd482b37cccff50ea6b2c6d2b069366fa683a9", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5, 14.875, 14.5625, 17.25, 17.75, 19.125], "runtime_probabilities": [0.01828356459736824, 0.009786486625671387, 0.0071599469520151615, 0.10521465539932251, 0.1734696328639984, 0.6860857009887695], "decide_seconds": 0.16156484999737586, "probabilities_f64_minus_runtime_max_abs": 1.3328662507028355e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.023242549566480405, "sagan": 0.05935192955146942, "socrates": 0.7230545233168743, "ibn_sina": 0.015974355128809743, "darwin": 0.03599876498607049, "hypatia": 0.1423778774502957}, "confidence": 0.4682269991158722, "choice": "socrates"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.20658199806348, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "e4cf79a383e4270bd75dbad0aced9f5f9b58e7bfd06ab3e39727d6af5a53f92e", "compiled_ids_sha256": "7d46106b95b4889d5ddfa6c63eb0979a5cbc2e457659e063755bcdfe321748e4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.8125, 16.75, 19.25, 15.4375, 16.25, 17.625], "runtime_probabilities": [0.02324254997074604, 0.059351928532123566, 0.7230545282363892, 0.01597435399889946, 0.035998761653900146, 0.14237786829471588], "decide_seconds": 0.1616736799987848, "probabilities_f64_minus_runtime_max_abs": 9.155579822062165e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.11675241816208425, "ibn_sina": 0.02773112999323369, "hypatia": 0.11675241816208425, "einstein": 0.03345012784846391, "sagan": 0.67186377798567, "darwin": 0.03345012784846391}, "confidence": 0.38863061886202344, "choice": "sagan"}}, "usage": {"input_tokens": 297, "output_tokens": 1}, "latency_ms": 165.27919199870666, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 297, "compiled_prompt_sha256": "1897c93ff7a27e62ef324efcaada8d4989f2f3b41b3a912beb3f9588c66e9713", "compiled_ids_sha256": "cdb2a8f6114e92846e6ae309493852e0a849d99637df8360978b8f3da0b230dd", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.25, 15.8125, 17.25, 16.0, 19.0, 16.0], "runtime_probabilities": [0.11675241589546204, 0.02773113176226616, 0.11675241589546204, 0.03345012664794922, 0.6718637943267822, 0.03345012664794922], "decide_seconds": 0.1616062899993267, "probabilities_f64_minus_runtime_max_abs": 1.6341112241136102e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:36Z"} {"id": "c-socrates-01-main-02", "uid": "c099", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "prompt_sha256_by_order": ["d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755"], "request_sha256": "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "compiled_prompt_sha256_by_order": ["241005cdbb0ac4de9754451ae5ce2a6d308deccb51feb8a49be378b2807af5d6", "eaec06d2423a7ba9b0c3f7fff3dca302237d8a5f27b823252a18121d4e1a7ba3", "a689d1686bffde1b551f8fd49c12f38a50a6ff9ad72e67027af2c3be66b3a033", "f6b1cf8376e48fa3129bc369507669dded1555c084c8a6ba61dcc3bdd8391cd4", "8067d15b1b4637576aecaa56150d4addfd6b218c11de084a93d142545a8513c0", "df18e9943e030e2e856eb69266f0acd5e414c0fd09095675a3c1f595e96c8d9d"], "compiled_tokens_by_order": [368, 368, 368, 368, 368, 368], "choice": "socrates", "probs": {"darwin": 0.0034495023732042636, "einstein": 0.0036719761622999447, "hypatia": 0.015459597084094763, "ibn_sina": 0.0175180185208784, "sagan": 0.0034495023732042636, "socrates": 0.9564514034863184}, "confidence": 0.8673926477126659, "seconds": 0.17110580233384098, "seconds_all_orders": 1.026634814003046, "seconds_by_order": [0.17149535499993362, 0.17141977500068606, 0.17097560800175415, 0.17076547000033315, 0.17085530899930745, 0.17112329700103146], "argmax_stable": true, "spread": {"darwin": 0.0019981922632145727, "einstein": 0.003896838602763575, "hypatia": 0.018341222003548603, "ibn_sina": 0.02881510949005026, "sagan": 0.0019981922632145727, "socrates": 0.031002497190860567}, "spread_2dp": {"darwin": 0.0, "einstein": 0.01, "hypatia": 0.02, "ibn_sina": 0.03, "sagan": 0.0, "socrates": 0.030000000000000027}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0034495023732042636, "einstein": 0.0036719761622999447, "hypatia": 0.015459597084094763, "ibn_sina": 0.0175180185208784, "sagan": 0.0034495023732042636, "socrates": 0.9564514034863184}, {"ibn_sina": 0.017669077377725835, "hypatia": 0.010716837158425618, "sagan": 0.0019824155401441876, "socrates": 0.96469893761631, "darwin": 0.0030704252530584686, "einstein": 0.0018623070543358825}, {"hypatia": 0.004260989095446691, "darwin": 0.003117406217000013, "sagan": 0.0016686273049854085, "ibn_sina": 0.007960575652231062, "einstein": 0.0035324840323744723, "socrates": 0.9794599176979623}, {"ibn_sina": 0.01293538620597098, "einstein": 0.005065570021495687, "darwin": 0.0014513101099896907, "sagan": 0.0014513101099896907, "hypatia": 0.013769646940218622, "socrates": 0.9653267766123352}, {"einstein": 0.00563973397569408, "sagan": 0.0019490399089940496, "socrates": 0.9484574205071017, "ibn_sina": 0.03677568514228132, "darwin": 0.0025026167812871223, "hypatia": 0.004675503684641707}, {"socrates": 0.9610704715231082, "ibn_sina": 0.01002967037706389, "hypatia": 0.022602211098995295, "einstein": 0.0017428953729305052, "sagan": 0.0026994491609440034, "darwin": 0.001855302466958099}], "prompt_tokens": 368, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0034495023732042636, "einstein": 0.0036719761622999447, "hypatia": 0.015459597084094763, "ibn_sina": 0.0175180185208784, "sagan": 0.0034495023732042636, "socrates": 0.9564514034863184}, "confidence": 0.8673926477126659, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 171.4683949976461, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "241005cdbb0ac4de9754451ae5ce2a6d308deccb51feb8a49be378b2807af5d6", "compiled_ids_sha256": "7479577962a5718990d77f717d29e7cc80e2c6bc96a00d11d0d6eae9329107e2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.125, 14.1875, 15.625, 15.75, 14.125, 19.75], "runtime_probabilities": [0.003449502633884549, 0.0036719765048474073, 0.015459597110748291, 0.017518019303679466, 0.003449502633884549, 0.956451416015625], "decide_seconds": 0.16718314700119663, "probabilities_f64_minus_runtime_max_abs": 1.2529306592234946e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.017669077377725835, "hypatia": 0.010716837158425618, "sagan": 0.0019824155401441876, "socrates": 0.96469893761631, "darwin": 0.0030704252530584686, "einstein": 0.0018623070543358825}, "confidence": 0.8903860430986874, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 171.3945949995832, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "eaec06d2423a7ba9b0c3f7fff3dca302237d8a5f27b823252a18121d4e1a7ba3", "compiled_ids_sha256": "8d305155a1c25196f54bb00326f4df85877094db1e198ad74a90f597c034047f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.0, 15.5, 13.8125, 20.0, 14.25, 13.75], "runtime_probabilities": [0.017669077962636948, 0.010716837830841541, 0.001982415560632944, 0.9646989107131958, 0.0030704252421855927, 0.0018623070791363716], "decide_seconds": 0.167052437998791, "probabilities_f64_minus_runtime_max_abs": 2.6903114180854004e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.004260989095446691, "darwin": 0.003117406217000013, "sagan": 0.0016686273049854085, "ibn_sina": 0.007960575652231062, "einstein": 0.0035324840323744723, "socrates": 0.9794599176979623}, "confidence": 0.9270738221858087, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 170.9506990009686, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "a689d1686bffde1b551f8fd49c12f38a50a6ff9ad72e67027af2c3be66b3a033", "compiled_ids_sha256": "3681413d94547e23758d234b431d9df1ca4568d2e26ae808ceda1258c6f3521f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 14.0, 13.375, 14.9375, 14.125, 19.75], "runtime_probabilities": [0.004260989371687174, 0.0031174062751233578, 0.0016686274902895093, 0.007960576564073563, 0.0035324841737747192, 0.9794599413871765], "decide_seconds": 0.16671159200268448, "probabilities_f64_minus_runtime_max_abs": 2.3689214256883417e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01293538620597098, "einstein": 0.005065570021495687, "darwin": 0.0014513101099896907, "sagan": 0.0014513101099896907, "hypatia": 0.013769646940218622, "socrates": 0.9653267766123352}, "confidence": 0.8911378808077437, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 170.74184000011883, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "f6b1cf8376e48fa3129bc369507669dded1555c084c8a6ba61dcc3bdd8391cd4", "compiled_ids_sha256": "51dfe3e0726a0d00ea433504f874c4f7735dcadf85dea872d59e972a8b78c724", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.5625, 14.625, 13.375, 13.375, 15.625, 19.875], "runtime_probabilities": [0.012935386970639229, 0.005065570119768381, 0.0014513101195916533, 0.0014513101195916533, 0.013769647106528282, 0.9653268456459045], "decide_seconds": 0.1665618719998747, "probabilities_f64_minus_runtime_max_abs": 6.903356930543225e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.00563973397569408, "sagan": 0.0019490399089940496, "socrates": 0.9484574205071017, "ibn_sina": 0.03677568514228132, "darwin": 0.0025026167812871223, "hypatia": 0.004675503684641707}, "confidence": 0.8587418765745062, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 170.8311889997276, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "8067d15b1b4637576aecaa56150d4addfd6b218c11de084a93d142545a8513c0", "compiled_ids_sha256": "ba3e8f90ccb2f79a64382c745df6f4d594567d0e08baaa8be0129ceb828f9d22", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.875, 13.8125, 20.0, 16.75, 14.0625, 14.6875], "runtime_probabilities": [0.005639733746647835, 0.001949039869941771, 0.9484574198722839, 0.03677568584680557, 0.0025026167277246714, 0.004675503820180893], "decide_seconds": 0.1666757319981116, "probabilities_f64_minus_runtime_max_abs": 7.045242514380767e-10}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9610704715231082, "ibn_sina": 0.01002967037706389, "hypatia": 0.022602211098995295, "einstein": 0.0017428953729305052, "sagan": 0.0026994491609440034, "darwin": 0.001855302466958099}, "confidence": 0.8835317246360244, "choice": "socrates"}}, "usage": {"input_tokens": 368, "output_tokens": 1}, "latency_ms": 171.09964699920965, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 368, "compiled_prompt_sha256": "df18e9943e030e2e856eb69266f0acd5e414c0fd09095675a3c1f595e96c8d9d", "compiled_ids_sha256": "6b1036d66a025f372cfece7382b11c77c1ac584a2277666034702f986e836522", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 15.1875, 16.0, 13.4375, 13.875, 13.5], "runtime_probabilities": [0.9610705375671387, 0.010029671713709831, 0.02260221354663372, 0.0017428954597562551, 0.002699449425563216, 0.0018553026020526886], "decide_seconds": 0.16688276999775553, "probabilities_f64_minus_runtime_max_abs": 6.604403046672758e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:37Z"} {"id": "c-socrates-01-main-02", "uid": "c100", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "prompt_sha256_by_order": ["b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2"], "request_sha256": "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "compiled_prompt_sha256_by_order": ["5a8c5648b9712c46350b863549c1ceed5801044f634662c32f1ac342b250aa4c", "2715c49d2ec43081a62bb93258e0ddb4d67c44598ed54685cf64c50a9a589424", "3a9ec945decbf700df228adc3d5522a7edff7ccf4ee4e66eb8181b8759d75a71", "80caaf9373c0b7ea823fa5b533d765c439fe9f6ac78199808a18c3b81c553d2e", "f5e6e09cffc1d28b199c0e08bd0888ac630076a3722af6e42e376e39a05786e7", "27e95771a2772083c0cfe4b58b1285d9da010e45929422d7a7cdbe0438f82348"], "compiled_tokens_by_order": [298, 298, 298, 298, 298, 298], "choice": "socrates", "probs": {"darwin": 0.0422699676061954, "einstein": 0.022625483244343592, "hypatia": 0.14753668374642429, "ibn_sina": 0.014608109916429404, "sagan": 0.18944085182422998, "socrates": 0.5835189036623774}, "confidence": 0.33416217939508097, "seconds": 0.1653656446669629, "seconds_all_orders": 0.9921938680017774, "seconds_by_order": [0.1657056680014648, 0.16603180699894438, 0.16505969400168397, 0.1650701239996124, 0.16532259099767543, 0.16500398400239646], "argmax_stable": false, "spread": {"darwin": 0.0813195460629537, "einstein": 0.18367566929000767, "hypatia": 0.1583610392849262, "ibn_sina": 0.07303305508066237, "sagan": 0.5064556152949488, "socrates": 0.5522581464628528}, "spread_2dp": {"darwin": 0.08, "einstein": 0.19, "hypatia": 0.16, "ibn_sina": 0.08, "sagan": 0.5, "socrates": 0.56}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0422699676061954, "einstein": 0.022625483244343592, "hypatia": 0.14753668374642429, "ibn_sina": 0.014608109916429404, "sagan": 0.18944085182422998, "socrates": 0.5835189036623774}, {"ibn_sina": 0.0591062697988774, "hypatia": 0.20630115253435127, "sagan": 0.0669761781891484, "socrates": 0.38542129423591, "darwin": 0.07589395270736163, "einstein": 0.20630115253435127}, {"hypatia": 0.1999657827048429, "darwin": 0.027062425839990863, "sagan": 0.06491938474237086, "ibn_sina": 0.04461839711913771, "einstein": 0.04749603649911868, "socrates": 0.6159379730945389}, {"ibn_sina": 0.048904348721381864, "einstein": 0.02617661155739771, "darwin": 0.0179909044784753, "sagan": 0.06279442674481496, "hypatia": 0.248356775547559, "socrates": 0.5957769329503712}, {"einstein": 0.08764116499709176, "sagan": 0.026729045153875825, "socrates": 0.3927804512791611, "ibn_sina": 0.08764116499709176, "darwin": 0.09931045054142901, "hypatia": 0.3058977230313505}, {"socrates": 0.06367982663168614, "ibn_sina": 0.03202022771012669, "hypatia": 0.22226443441629673, "einstein": 0.056197249759587044, "sagan": 0.5331846604488246, "darwin": 0.09265360103347883}], "prompt_tokens": 298, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0422699676061954, "einstein": 0.022625483244343592, "hypatia": 0.14753668374642429, "ibn_sina": 0.014608109916429404, "sagan": 0.18944085182422998, "socrates": 0.5835189036623774}, "confidence": 0.33416217939508097, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.67893799947342, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "5a8c5648b9712c46350b863549c1ceed5801044f634662c32f1ac342b250aa4c", "compiled_ids_sha256": "b7bb323dfd3806fa1b596523b49ddfd4bc24913a34b61cf4015eef0afa9175a5", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 15.5, 17.375, 15.0625, 17.625, 18.75], "runtime_probabilities": [0.04226996749639511, 0.022625481709837914, 0.14753668010234833, 0.014608110301196575, 0.18944084644317627, 0.5835188627243042], "decide_seconds": 0.16190910799923586, "probabilities_f64_minus_runtime_max_abs": 4.0938073153640175e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0591062697988774, "hypatia": 0.20630115253435127, "sagan": 0.0669761781891484, "socrates": 0.38542129423591, "darwin": 0.07589395270736163, "einstein": 0.20630115253435127}, "confidence": 0.12786566467870109, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 166.00823699991452, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "2715c49d2ec43081a62bb93258e0ddb4d67c44598ed54685cf64c50a9a589424", "compiled_ids_sha256": "cffb8432c20632ab08338fbc36efe71f8f4f981a448d75e3144a7838f7d8db85", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 17.375, 16.25, 18.0, 16.375, 17.375], "runtime_probabilities": [0.05910627171397209, 0.20630115270614624, 0.06697617471218109, 0.3854212760925293, 0.07589394599199295, 0.20630115270614624], "decide_seconds": 0.1621793259982951, "probabilities_f64_minus_runtime_max_abs": 1.814338068673038e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.1999657827048429, "darwin": 0.027062425839990863, "sagan": 0.06491938474237086, "ibn_sina": 0.04461839711913771, "einstein": 0.04749603649911868, "socrates": 0.6159379730945389}, "confidence": 0.3419646321250944, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.03626400299254, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "3a9ec945decbf700df228adc3d5522a7edff7ccf4ee4e66eb8181b8759d75a71", "compiled_ids_sha256": "3cc932c6fa6a6ebf003196cf9406c5ed21a22ecce851734ffd2fe25380151d32", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [17.375, 15.375, 16.25, 15.875, 15.9375, 18.5], "runtime_probabilities": [0.19996577501296997, 0.027062425389885902, 0.0649193823337555, 0.0446183979511261, 0.04749603569507599, 0.6159379482269287], "decide_seconds": 0.16139326199845527, "probabilities_f64_minus_runtime_max_abs": 2.4867610193268774e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.048904348721381864, "einstein": 0.02617661155739771, "darwin": 0.0179909044784753, "sagan": 0.06279442674481496, "hypatia": 0.248356775547559, "socrates": 0.5957769329503712}, "confidence": 0.36178959142578326, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.04776399960974, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "80caaf9373c0b7ea823fa5b533d765c439fe9f6ac78199808a18c3b81c553d2e", "compiled_ids_sha256": "ce647e2cd3d97f071c8cf32eaa7ddb768691ca7251af97c552b35a416866fe9b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.125, 15.5, 15.125, 16.375, 17.75, 18.625], "runtime_probabilities": [0.04890434816479683, 0.026176610961556435, 0.017990903928875923, 0.06279442459344864, 0.24835675954818726, 0.5957769155502319], "decide_seconds": 0.1614647509995848, "probabilities_f64_minus_runtime_max_abs": 1.740013921835981e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.08764116499709176, "sagan": 0.026729045153875825, "socrates": 0.3927804512791611, "ibn_sina": 0.08764116499709176, "darwin": 0.09931045054142901, "hypatia": 0.3058977230313505}, "confidence": 0.17271899442353544, "choice": "socrates"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 165.2999420002743, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "f5e6e09cffc1d28b199c0e08bd0888ac630076a3722af6e42e376e39a05786e7", "compiled_ids_sha256": "04afb2fece8b1c7c6f4d7c3408d35bcd36749267ddc32787bc5a29fea7b54184", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.75, 15.5625, 18.25, 16.75, 16.875, 18.0], "runtime_probabilities": [0.0876411646604538, 0.026729045435786247, 0.39278045296669006, 0.0876411646604538, 0.09931045025587082, 0.30589771270751953], "decide_seconds": 0.16163311000127578, "probabilities_f64_minus_runtime_max_abs": 1.0323830956782132e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.06367982663168614, "ibn_sina": 0.03202022771012669, "hypatia": 0.22226443441629673, "einstein": 0.056197249759587044, "sagan": 0.5331846604488246, "darwin": 0.09265360103347883}, "confidence": 0.25362025494011853, "choice": "sagan"}}, "usage": {"input_tokens": 298, "output_tokens": 1}, "latency_ms": 164.98102500190726, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 298, "compiled_prompt_sha256": "27e95771a2772083c0cfe4b58b1285d9da010e45929422d7a7cdbe0438f82348", "compiled_ids_sha256": "e6b41d31c07ac2c04e603e631bcabb0dde14772c1b815d2a7205d21c80798ceb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 15.6875, 17.625, 16.25, 18.5, 16.75], "runtime_probabilities": [0.06367982923984528, 0.03202022984623909, 0.22226443886756897, 0.05619725212454796, 0.5331847071647644, 0.09265359491109848], "decide_seconds": 0.16136269200069364, "probabilities_f64_minus_runtime_max_abs": 4.671593978322619e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:38Z"} {"id": "c-socrates-01-main-03", "uid": "c101", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "prompt_sha256_by_order": ["087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b"], "request_sha256": "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "compiled_prompt_sha256_by_order": ["aa8a76f665cd82962eddc0c1cfe089528415b48ad3760ba5f5715122b5f5c9b1", "752ba78f33d5a48f1c426bacf0099db95d6283dcafeef13c404c42674ae7e498", "0ea387a6e148bc3af33e032902215e3d7192a6bcce27aa972b4003b859b57a3c", "7d6e81fb87cb2bd9f961687c450633f3ea84686727cd1830a704b84e559427e0", "c362ea45c5237f201996344a7fb1da3574abbfb717ef1a14d8d12717b123b784", "f533ca7dc314f0448dd4327d2fc5474b352157a292ad23d17cc8b3beb05741a7"], "compiled_tokens_by_order": [340, 340, 340, 340, 340, 340], "choice": "sagan", "probs": {"darwin": 0.005791938695948941, "einstein": 0.007436996497495692, "hypatia": 0.010820766054401866, "ibn_sina": 0.0024144392626473813, "sagan": 0.9150393508685083, "socrates": 0.05849650862099779}, "confidence": 0.7895260322699845, "seconds": 0.17064713933298967, "seconds_all_orders": 1.023882835997938, "seconds_by_order": [0.17053071199916303, 0.17080215999885695, 0.1708181499998318, 0.17057228100020438, 0.17061673100033659, 0.1705428019995452], "argmax_stable": true, "spread": {"darwin": 0.009925997734714264, "einstein": 0.06063731801733421, "hypatia": 0.0203801881602685, "ibn_sina": 0.004784111371050681, "sagan": 0.09103622853530535, "socrates": 0.05231403833027454}, "spread_2dp": {"darwin": 0.01, "einstein": 0.06, "hypatia": 0.02, "ibn_sina": 0.01, "sagan": 0.08999999999999997, "socrates": 0.049999999999999996}, "choices_by_order": ["sagan", "sagan", "sagan", "sagan", "sagan", "sagan"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.005791938695948941, "einstein": 0.007436996497495692, "hypatia": 0.010820766054401866, "ibn_sina": 0.0024144392626473813, "sagan": 0.9150393508685083, "socrates": 0.05849650862099779}, {"ibn_sina": 0.0071985506336980625, "hypatia": 0.0043661416648316815, "sagan": 0.8857012506690346, "socrates": 0.025125410508260507, "darwin": 0.013448663120777568, "einstein": 0.06415998340339751}, {"hypatia": 0.004710812498286693, "darwin": 0.0036687844625682376, "sagan": 0.9556200512244927, "ibn_sina": 0.003237690924466959, "einstein": 0.007296249128530784, "socrates": 0.025466411761654522}, {"ibn_sina": 0.0037008262498440013, "einstein": 0.004193585540079834, "darwin": 0.004193585540079834, "sagan": 0.9639660782833718, "hypatia": 0.007359971831478481, "socrates": 0.01658595255514607}, {"einstein": 0.01500940775428443, "sagan": 0.9285991759030618, "socrates": 0.021838559420987703, "ibn_sina": 0.003146142779692004, "darwin": 0.006660384316873818, "hypatia": 0.024746329825100184}, {"socrates": 0.00618247029072325, "ibn_sina": 0.0025772370519932686, "hypatia": 0.007457482680816908, "einstein": 0.0035226653860633036, "sagan": 0.97673747920434, "darwin": 0.0035226653860633036}], "prompt_tokens": 340, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.005791938695948941, "einstein": 0.007436996497495692, "hypatia": 0.010820766054401866, "ibn_sina": 0.0024144392626473813, "sagan": 0.9150393508685083, "socrates": 0.05849650862099779}, "confidence": 0.7895260322699845, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.50356199979433, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "aa8a76f665cd82962eddc0c1cfe089528415b48ad3760ba5f5715122b5f5c9b1", "compiled_ids_sha256": "86a40936a666d5fd48b556179b1723278e9807945ccc3cb11a9396580f0dd5a4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.4375, 14.6875, 15.0625, 13.5625, 19.5, 16.75], "runtime_probabilities": [0.005791937932372093, 0.007436996325850487, 0.010820765048265457, 0.002414439106360078, 0.9150393009185791, 0.05849650129675865], "decide_seconds": 0.16644292300043162, "probabilities_f64_minus_runtime_max_abs": 4.994992919282737e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0071985506336980625, "hypatia": 0.0043661416648316815, "sagan": 0.8857012506690346, "socrates": 0.025125410508260507, "darwin": 0.013448663120777568, "einstein": 0.06415998340339751}, "confidence": 0.7245951572848797, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.7781500008423, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "752ba78f33d5a48f1c426bacf0099db95d6283dcafeef13c404c42674ae7e498", "compiled_ids_sha256": "5459472ab449027ca2803ae13899f655ebd87eb091b0d2dd09405639570009c3", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.0625, 19.375, 15.8125, 15.1875, 16.75], "runtime_probabilities": [0.00719855073839426, 0.004366141278296709, 0.8857011795043945, 0.0251254104077816, 0.013448662124574184, 0.06415998190641403], "decide_seconds": 0.16684258000168484, "probabilities_f64_minus_runtime_max_abs": 7.116464006973189e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.004710812498286693, "darwin": 0.0036687844625682376, "sagan": 0.9556200512244927, "ibn_sina": 0.003237690924466959, "einstein": 0.007296249128530784, "socrates": 0.025466411761654522}, "confidence": 0.8676563376948031, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.7925199989404, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "0ea387a6e148bc3af33e032902215e3d7192a6bcce27aa972b4003b859b57a3c", "compiled_ids_sha256": "9fab0fceceb800bddcf101132b2871a0786b128ec068cd02eec060322221956c", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.4375, 14.1875, 19.75, 14.0625, 14.875, 16.125], "runtime_probabilities": [0.004710812587291002, 0.0036687846295535564, 0.9556201100349426, 0.0032376910094171762, 0.007296248804777861, 0.02546641230583191], "decide_seconds": 0.1666439309992711, "probabilities_f64_minus_runtime_max_abs": 5.881044995437179e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0037008262498440013, "einstein": 0.004193585540079834, "darwin": 0.004193585540079834, "sagan": 0.9639660782833718, "hypatia": 0.007359971831478481, "socrates": 0.01658595255514607}, "confidence": 0.8849451833626264, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.54830200140714, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "7d6e81fb87cb2bd9f961687c450633f3ea84686727cd1830a704b84e559427e0", "compiled_ids_sha256": "221af17b634b5509eba3321d659ff40244dbf72a25ae37b973e755200ab33acf", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.1875, 14.3125, 14.3125, 19.75, 14.875, 15.6875], "runtime_probabilities": [0.003700826782733202, 0.004193585831671953, 0.004193585831671953, 0.9639661312103271, 0.007359972223639488, 0.016585953533649445], "decide_seconds": 0.16655335299947183, "probabilities_f64_minus_runtime_max_abs": 5.292695537395531e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.01500940775428443, "sagan": 0.9285991759030618, "socrates": 0.021838559420987703, "ibn_sina": 0.003146142779692004, "darwin": 0.006660384316873818, "hypatia": 0.024746329825100184}, "confidence": 0.7999891826618956, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.59275100109517, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "c362ea45c5237f201996344a7fb1da3574abbfb717ef1a14d8d12717b123b784", "compiled_ids_sha256": "a01d19e0e7fbf06b82ca23cb2e27420bcad2a1a648449f1f290a7c6d25067685", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.375, 19.5, 15.75, 13.8125, 14.5625, 15.875], "runtime_probabilities": [0.015009406954050064, 0.9285991191864014, 0.021838558837771416, 0.0031461426988244057, 0.00666038366034627, 0.024746328592300415], "decide_seconds": 0.1666539420002664, "probabilities_f64_minus_runtime_max_abs": 5.671666047923907e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.00618247029072325, "ibn_sina": 0.0025772370519932686, "hypatia": 0.007457482680816908, "einstein": 0.0035226653860633036, "sagan": 0.97673747920434, "darwin": 0.0035226653860633036}, "confidence": 0.9184468222563352, "choice": "sagan"}}, "usage": {"input_tokens": 340, "output_tokens": 1}, "latency_ms": 170.51892199742724, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 340, "compiled_prompt_sha256": "f533ca7dc314f0448dd4327d2fc5474b352157a292ad23d17cc8b3beb05741a7", "compiled_ids_sha256": "53702ffe5108955515bdfb2ddaecf0b23e58a9d927d626522e8e7901770cd8d8", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.3125, 13.4375, 14.5, 13.75, 19.375, 13.75], "runtime_probabilities": [0.006182469893246889, 0.0025772370863705873, 0.007457482162863016, 0.0035226652398705482, 0.9767374396324158, 0.0035226652398705482], "decide_seconds": 0.16651869199995417, "probabilities_f64_minus_runtime_max_abs": 3.957192418191369e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:39Z"} {"id": "c-socrates-01-main-03", "uid": "c102", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "prompt_sha256_by_order": ["da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb"], "request_sha256": "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "compiled_prompt_sha256_by_order": ["302cd7d856db5d188e298657e794f2abd5a5c6239133269be866a306600f9f17", "878c56ee00af1549d43d0014299394f15d4be3e597a4917593617077f919ecb1", "b403e3a6bbc5f872856896adacda3ce13eb1b0014257ef2496f6e08ef82a92a5", "fdd319adf77f2aea4125d2a2e8fc438e60c0b15948b8a31576a379579411b3c5", "cb5d2bf455d73adc8c5ac8b875b4d226401ca3180f9c99dfffb61606b9034de0", "d02043f78697ca10320140700161350c84ec358c36ac6c5f0abd5a2ade19c74b"], "compiled_tokens_by_order": [380, 380, 380, 380, 380, 380], "choice": "hypatia", "probs": {"darwin": 0.008613938339646177, "einstein": 0.0040689363590223245, "hypatia": 0.8786455978072959, "ibn_sina": 0.009760870904383092, "sagan": 0.006302091916067829, "socrates": 0.09260856467358472}, "confidence": 0.7351801819093946, "seconds": 0.17155696733364798, "seconds_all_orders": 1.0293418040018878, "seconds_by_order": [0.17175107300136006, 0.17174602199884248, 0.17172192300131428, 0.17124535699986154, 0.17152320400055032, 0.17135422499995911], "argmax_stable": true, "spread": {"darwin": 0.01733191926896028, "einstein": 0.00402302939214122, "hypatia": 0.417721209392399, "ibn_sina": 0.018104122387583493, "sagan": 0.011046187679708129, "socrates": 0.4485281899206637}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.42000000000000004, "ibn_sina": 0.019999999999999997, "sagan": 0.01, "socrates": 0.44999999999999996}, "choices_by_order": ["hypatia", "hypatia", "hypatia", "hypatia", "hypatia", "hypatia"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.008613938339646177, "einstein": 0.0040689363590223245, "hypatia": 0.8786455978072959, "ibn_sina": 0.009760870904383092, "sagan": 0.006302091916067829, "socrates": 0.09260856467358472}, {"ibn_sina": 0.01823552631718491, "hypatia": 0.5329201959935899, "sagan": 0.0063020292635931575, "socrates": 0.41503866595437433, "darwin": 0.019411616720094137, "einstein": 0.008091965751163544}, {"hypatia": 0.479219824331206, "darwin": 0.017455580825276335, "sagan": 0.005666997384099061, "ibn_sina": 0.012770775744113486, "einstein": 0.005666997384099061, "socrates": 0.479219824331206}, {"ibn_sina": 0.026744304665283283, "einstein": 0.005266265833812273, "darwin": 0.007198131949961873, "sagan": 0.005266265833812273, "hypatia": 0.5371737188395276, "socrates": 0.41835131287760274}, {"einstein": 0.0048072631332348225, "sagan": 0.002573142531961948, "socrates": 0.11646932423072716, "ibn_sina": 0.008981149914978916, "darwin": 0.006570749643711378, "hypatia": 0.8605983705453858}, {"socrates": 0.03069163441054235, "ibn_sina": 0.02708527230256241, "hypatia": 0.896941033723605, "einstein": 0.007760060438948576, "sagan": 0.013619330211670078, "darwin": 0.02390266891267166}], "prompt_tokens": 380, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.008613938339646177, "einstein": 0.0040689363590223245, "hypatia": 0.8786455978072959, "ibn_sina": 0.009760870904383092, "sagan": 0.006302091916067829, "socrates": 0.09260856467358472}, "confidence": 0.7351801819093946, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.72280300292186, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "302cd7d856db5d188e298657e794f2abd5a5c6239133269be866a306600f9f17", "compiled_ids_sha256": "064b264cbea6255d0964374c8390710e5279938b074c351063f0f26916c2824f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.125, 14.375, 19.75, 15.25, 14.8125, 17.5], "runtime_probabilities": [0.008613938465714455, 0.004068936686962843, 0.8786455988883972, 0.009760871529579163, 0.006302091758698225, 0.09260856360197067], "decide_seconds": 0.16722087699963595, "probabilities_f64_minus_runtime_max_abs": 1.0811013195777264e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.01823552631718491, "hypatia": 0.5329201959935899, "sagan": 0.0063020292635931575, "socrates": 0.41503866595437433, "darwin": 0.019411616720094137, "einstein": 0.008091965751163544}, "confidence": 0.486069688724173, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.72077299983357, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "878c56ee00af1549d43d0014299394f15d4be3e597a4917593617077f919ecb1", "compiled_ids_sha256": "bd9a0dfc1c36995835124b983ac3cccc71f6ddb227d9fafbe99a3dcd51c84a6f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 19.125, 14.6875, 18.875, 15.8125, 14.9375], "runtime_probabilities": [0.018235526978969574, 0.5329201817512512, 0.0063020288944244385, 0.41503867506980896, 0.019411617890000343, 0.008091965690255165], "decide_seconds": 0.1672705870005302, "probabilities_f64_minus_runtime_max_abs": 1.4242338641778929e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.479219824331206, "darwin": 0.017455580825276335, "sagan": 0.005666997384099061, "ibn_sina": 0.012770775744113486, "einstein": 0.005666997384099061, "socrates": 0.479219824331206}, "confidence": 0.5032781162749227, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.6970930028765, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "b403e3a6bbc5f872856896adacda3ce13eb1b0014257ef2496f6e08ef82a92a5", "compiled_ids_sha256": "0a148a20c59656a2bb852819d80b63ec2e6a27360847d4a5408ab470e318ce97", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.125, 15.8125, 14.6875, 15.5, 14.6875, 19.125], "runtime_probabilities": [0.47921982407569885, 0.017455581575632095, 0.005666997283697128, 0.012770775705575943, 0.005666997283697128, 0.47921982407569885], "decide_seconds": 0.16713654799968936, "probabilities_f64_minus_runtime_max_abs": 7.503557601651334e-10}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.026744304665283283, "einstein": 0.005266265833812273, "darwin": 0.007198131949961873, "sagan": 0.005266265833812273, "hypatia": 0.5371737188395276, "socrates": 0.41835131287760274}, "confidence": 0.505508747995292, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.21999699884327, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "fdd319adf77f2aea4125d2a2e8fc438e60c0b15948b8a31576a379579411b3c5", "compiled_ids_sha256": "eb004f6cadad59ebf9a174777f1c5594f626f7a6913fd626368744ce692b45ab", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 14.625, 14.9375, 14.625, 19.25, 19.0], "runtime_probabilities": [0.026744304224848747, 0.005266265943646431, 0.007198132108896971, 0.005266265943646431, 0.5371737480163574, 0.41835132241249084], "decide_seconds": 0.1669194089990924, "probabilities_f64_minus_runtime_max_abs": 2.9176829841048857e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0048072631332348225, "sagan": 0.002573142531961948, "socrates": 0.11646932423072716, "ibn_sina": 0.008981149914978916, "darwin": 0.006570749643711378, "hypatia": 0.8605983705453858}, "confidence": 0.7231944474560934, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.4983450001455, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "cb5d2bf455d73adc8c5ac8b875b4d226401ca3180f9c99dfffb61606b9034de0", "compiled_ids_sha256": "01dd6d124fff090dfc66e3f02e6dd3d270b911b630deeeabbe1f0f04569ca867", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.0625, 17.875, 15.3125, 15.0, 19.875], "runtime_probabilities": [0.004807263612747192, 0.002573142759501934, 0.11646932363510132, 0.008981150574982166, 0.0065707494504749775, 0.8605983853340149], "decide_seconds": 0.16706795799836982, "probabilities_f64_minus_runtime_max_abs": 1.4788629099093953e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.03069163441054235, "ibn_sina": 0.02708527230256241, "hypatia": 0.896941033723605, "einstein": 0.007760060438948576, "sagan": 0.013619330211670078, "darwin": 0.02390266891267166}, "confidence": 0.7278171032340033, "choice": "hypatia"}}, "usage": {"input_tokens": 380, "output_tokens": 1}, "latency_ms": 171.3297460009926, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 380, "compiled_prompt_sha256": "d02043f78697ca10320140700161350c84ec358c36ac6c5f0abd5a2ade19c74b", "compiled_ids_sha256": "9e231a171471ce8e90baadb5861bf4c2f421c5985f1eef02045a1f6bc829141f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.25, 16.125, 19.625, 14.875, 15.4375, 16.0], "runtime_probabilities": [0.030691631138324738, 0.02708527073264122, 0.8969409465789795, 0.007760059554129839, 0.01361932884901762, 0.023902665823698044], "decide_seconds": 0.16701187899889192, "probabilities_f64_minus_runtime_max_abs": 8.714462551573376e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:40Z"} {"id": "c-socrates-01-starter-01", "uid": "c103", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "prompt_sha256_by_order": ["ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c"], "request_sha256": "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "compiled_prompt_sha256_by_order": ["7ba348fbd49cae74d977fc56e0a7c6c7e8609ef11e84d162768dc1a823298c56", "3f5a56ceaad52ea0d53e09f9207d50a24b2c91cc0f25122104a03b375067966d", "a5c584d3428cacb1c6d335557737106622f032c16affb109e5eacf7e649ad131", "02d48ce06fd008951db9a3bc5e3a5bba5372afe6ad859080d27399c8b623e642", "116ea19dea447cdf35718566bd027424c4d502f6007db60def9a1e508f969933", "3c991d8fb09f07f5f4233d8c042b256cdf894c97b925bfc524776aecbcc8e0c7"], "compiled_tokens_by_order": [267, 267, 267, 267, 267, 267], "choice": "socrates", "probs": {"darwin": 0.0035588384968075716, "einstein": 0.002158544661279413, "hypatia": 0.002771626208138398, "ibn_sina": 0.0024459515438044234, "sagan": 0.002297758831258663, "socrates": 0.9867672802587114}, "confidence": 0.9489609054052881, "seconds": 0.16448747499998717, "seconds_all_orders": 0.986924849999923, "seconds_by_order": [0.16468092599825468, 0.16448253900307463, 0.1648064559994964, 0.16454463799891528, 0.1641900809991057, 0.16422021000107634], "argmax_stable": true, "spread": {"darwin": 0.0044386842242365035, "einstein": 0.005701942184238889, "hypatia": 0.006150040332109374, "ibn_sina": 0.00929435027847643, "sagan": 0.0020352742730289503, "socrates": 0.021118211342376014}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.01, "ibn_sina": 0.01, "sagan": 0.0, "socrates": 0.020000000000000018}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.0035588384968075716, "einstein": 0.002158544661279413, "hypatia": 0.002771626208138398, "ibn_sina": 0.0024459515438044234, "sagan": 0.002297758831258663, "socrates": 0.9867672802587114}, {"ibn_sina": 0.011455646596890724, "hypatia": 0.008921666540247772, "sagan": 0.0037191039328658563, "socrates": 0.9687269249790055, "darwin": 0.004775423977103099, "einstein": 0.002401233973887018}, {"hypatia": 0.004037833109690652, "darwin": 0.002300687955031587, "sagan": 0.001791777580981549, "ibn_sina": 0.0021612963184142954, "einstein": 0.0016832192652303974, "socrates": 0.9880251857706515}, {"ibn_sina": 0.0035646674178476446, "einstein": 0.0019080289742722306, "darwin": 0.0016838296598369062, "sagan": 0.0016838296598369062, "hypatia": 0.0027761657764088688, "socrates": 0.9883834785117974}, {"einstein": 0.007385161449469286, "sagan": 0.0037134923852068027, "socrates": 0.9672652671694214, "ibn_sina": 0.010745346504547122, "darwin": 0.0061225138840734095, "hypatia": 0.0047682186072819195}, {"socrates": 0.9854874311846279, "ibn_sina": 0.00402746188673066, "hypatia": 0.00402746188673066, "einstein": 0.001902438287730785, "sagan": 0.0027680313754892437, "darwin": 0.0017871753786908012}], "prompt_tokens": 267, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.0035588384968075716, "einstein": 0.002158544661279413, "hypatia": 0.002771626208138398, "ibn_sina": 0.0024459515438044234, "sagan": 0.002297758831258663, "socrates": 0.9867672802587114}, "confidence": 0.9489609054052881, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.6560369990766, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "7ba348fbd49cae74d977fc56e0a7c6c7e8609ef11e84d162768dc1a823298c56", "compiled_ids_sha256": "279db28cb15e54380a75c8b3d192304b55a16372b39c8bf6c63f78d6061d4052", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 13.25, 13.5, 13.375, 13.3125, 19.375], "runtime_probabilities": [0.003558838739991188, 0.002158544957637787, 0.002771626692265272, 0.0024459518026560545, 0.002297758823260665, 0.9867673516273499], "decide_seconds": 0.16130741300003137, "probabilities_f64_minus_runtime_max_abs": 7.136863844348085e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.011455646596890724, "hypatia": 0.008921666540247772, "sagan": 0.0037191039328658563, "socrates": 0.9687269249790055, "darwin": 0.004775423977103099, "einstein": 0.002401233973887018}, "confidence": 0.896810021557304, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.46093900231062, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "3f5a56ceaad52ea0d53e09f9207d50a24b2c91cc0f25122104a03b375067966d", "compiled_ids_sha256": "601c34aa442f08593a2e3803f078d24eeef568aa02631b4b4060bbc617bc38ee", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.8125, 13.9375, 19.5, 14.1875, 13.5], "runtime_probabilities": [0.01145564578473568, 0.008921666070818901, 0.003719103755429387, 0.9687268733978271, 0.004775423556566238, 0.002401233883574605], "decide_seconds": 0.16122055300002103, "probabilities_f64_minus_runtime_max_abs": 5.158117832859688e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.004037833109690652, "darwin": 0.002300687955031587, "sagan": 0.001791777580981549, "ibn_sina": 0.0021612963184142954, "einstein": 0.0016832192652303974, "socrates": 0.9880251857706515}, "confidence": 0.9534077241398935, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.78473600000143, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "a5c584d3428cacb1c6d335557737106622f032c16affb109e5eacf7e649ad131", "compiled_ids_sha256": "b2f8d83a099b8fa9ecac061d465635552b171fa65cde774a53500a42450b8c32", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 13.4375, 13.1875, 13.375, 13.125, 19.5], "runtime_probabilities": [0.00403783330693841, 0.0023006880655884743, 0.001791777671314776, 0.0021612965501844883, 0.0016832195688039064, 0.9880252480506897], "decide_seconds": 0.16156744000181789, "probabilities_f64_minus_runtime_max_abs": 6.228003823771644e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0035646674178476446, "einstein": 0.0019080289742722306, "darwin": 0.0016838296598369062, "sagan": 0.0016838296598369062, "hypatia": 0.0027761657764088688, "socrates": 0.9883834785117974}, "confidence": 0.9545476057436003, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.52235800170456, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "02d48ce06fd008951db9a3bc5e3a5bba5372afe6ad859080d27399c8b623e642", "compiled_ids_sha256": "6d65abb3870c492f6868afb827346392da2b104c5435416ed7602e8907480b96", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 13.125, 13.0, 13.0, 13.5, 19.375], "runtime_probabilities": [0.003564667422324419, 0.0019080289639532566, 0.0016838298179209232, 0.0016838298179209232, 0.002776165958493948, 0.9883834719657898], "decide_seconds": 0.16113096399931237, "probabilities_f64_minus_runtime_max_abs": 6.546007624841366e-09}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.007385161449469286, "sagan": 0.0037134923852068027, "socrates": 0.9672652671694214, "ibn_sina": 0.010745346504547122, "darwin": 0.0061225138840734095, "hypatia": 0.0047682186072819195}, "confidence": 0.8913794427273636, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.16834099800326, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "116ea19dea447cdf35718566bd027424c4d502f6007db60def9a1e508f969933", "compiled_ids_sha256": "dda3eb65361d6c9c41a6d01f4925e09bfac28339a36c3b7511c4dd62c2740d15", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 13.8125, 19.375, 14.875, 14.3125, 14.0625], "runtime_probabilities": [0.0073851607739925385, 0.003713492536917329, 0.967265248298645, 0.01074534747749567, 0.006122513674199581, 0.004768218379467726], "decide_seconds": 0.16093989499859163, "probabilities_f64_minus_runtime_max_abs": 1.8870776385959687e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9854874311846279, "ibn_sina": 0.00402746188673066, "hypatia": 0.00402746188673066, "einstein": 0.001902438287730785, "sagan": 0.0027680313754892437, "darwin": 0.0017871753786908012}, "confidence": 0.9451070543419113, "choice": "socrates"}}, "usage": {"input_tokens": 267, "output_tokens": 1}, "latency_ms": 164.19895000217366, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 267, "compiled_prompt_sha256": "3c991d8fb09f07f5f4233d8c042b256cdf894c97b925bfc524776aecbcc8e0c7", "compiled_ids_sha256": "cefc78669241588f0f8cbfa7a9289d44dc5a78227a0565f397573587aae36023", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 14.0, 14.0, 13.25, 13.625, 13.1875], "runtime_probabilities": [0.9854874610900879, 0.004027462098747492, 0.004027462098747492, 0.0019024383509531617, 0.0027680315542966127, 0.00178717530798167], "decide_seconds": 0.16095071499876212, "probabilities_f64_minus_runtime_max_abs": 2.990545999725924e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:41Z"} {"id": "c-socrates-01-starter-02", "uid": "c104", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "prompt_sha256_by_order": ["caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d"], "request_sha256": "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "compiled_prompt_sha256_by_order": ["09a3150ed8b79aaef3b11c5a10df28e718cebf9189825e56bd6872be31be436d", "15b4cd0fceefa103657844be6d1ba89db7dfa950d506eb26c21b6c0e921ad99b", "81f07b4c59c6b0e2a3ac361cc52b529a009e086ae13d368060b101bacf0501f0", "297cf3e7b0cb93d214dff7d93af8d03747074f169f88e60e5870669375322034", "57b0a5c48171094738510bfa2a4aae666897260d2743d7f7cecd03c9a8caecee", "513bcaa62245dda4f0599b7a24d1bf790b2259b02178ab05dab7f744b736f641"], "compiled_tokens_by_order": [315, 315, 315, 315, 315, 315], "choice": "socrates", "probs": {"darwin": 0.00953987287010779, "einstein": 0.005106325980343455, "hypatia": 0.045512188053175744, "ibn_sina": 0.016742998307491885, "sagan": 0.008961881191759143, "socrates": 0.914136733597122}, "confidence": 0.7741063348955545, "seconds": 0.16547196049941704, "seconds_all_orders": 0.9928317629965022, "seconds_by_order": [0.16536002199791255, 0.1653112020030676, 0.16535645099793328, 0.16594420700130286, 0.16550119999737944, 0.1653586809989065], "argmax_stable": true, "spread": {"darwin": 0.00825183909896745, "einstein": 0.007335757102935613, "hypatia": 0.04098693216300347, "ibn_sina": 0.01699220398709135, "sagan": 0.007502347716493637, "socrates": 0.06763883880482191}, "spread_2dp": {"darwin": 0.01, "einstein": 0.01, "hypatia": 0.05, "ibn_sina": 0.01, "sagan": 0.01, "socrates": 0.06999999999999995}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.00953987287010779, "einstein": 0.005106325980343455, "hypatia": 0.045512188053175744, "ibn_sina": 0.016742998307491885, "sagan": 0.008961881191759143, "socrates": 0.914136733597122}, {"ibn_sina": 0.008454291023160702, "hypatia": 0.004525255890172272, "sagan": 0.002275441559977619, "socrates": 0.977183179096655, "darwin": 0.004251084495801457, "einstein": 0.0033107479342328826}, {"hypatia": 0.019680873473475427, "darwin": 0.005297035235186961, "sagan": 0.0032128142757190123, "ibn_sina": 0.0153274796726719, "einstein": 0.008204208665190147, "socrates": 0.9482775886777565}, {"ibn_sina": 0.005772566874345978, "einstein": 0.0029026289103701707, "darwin": 0.0012880337711403391, "sagan": 0.0014595334752655058, "hypatia": 0.01778075774786658, "socrates": 0.9707964792210113}, {"einstein": 0.009231029817526272, "sagan": 0.0040962446508055255, "socrates": 0.9415906397967672, "ibn_sina": 0.022144089367137385, "darwin": 0.0033959060907084865, "hypatia": 0.019542090277055217}, {"socrates": 0.9817755724019439, "ibn_sina": 0.0051518853800460365, "hypatia": 0.007979396171705526, "einstein": 0.0018952727145906594, "sagan": 0.0018952727145906594, "darwin": 0.0013026006171232224}], "prompt_tokens": 315, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.00953987287010779, "einstein": 0.005106325980343455, "hypatia": 0.045512188053175744, "ibn_sina": 0.016742998307491885, "sagan": 0.008961881191759143, "socrates": 0.914136733597122}, "confidence": 0.7741063348955545, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.3350220003631, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "09a3150ed8b79aaef3b11c5a10df28e718cebf9189825e56bd6872be31be436d", "compiled_ids_sha256": "2fae6a509382acdcb1398799443b33dcf79f533516bd7b6f1ae007a1d10494dc", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0625, 14.4375, 16.625, 15.625, 15.0, 19.625], "runtime_probabilities": [0.009539873339235783, 0.005106326192617416, 0.04551218822598457, 0.016742998734116554, 0.008961881510913372, 0.9141367673873901], "decide_seconds": 0.16155868100031512, "probabilities_f64_minus_runtime_max_abs": 3.379026813910002e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.008454291023160702, "hypatia": 0.004525255890172272, "sagan": 0.002275441559977619, "socrates": 0.977183179096655, "darwin": 0.004251084495801457, "einstein": 0.0033107479342328826}, "confidence": 0.9200214629112553, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.28811200259952, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "15b4cd0fceefa103657844be6d1ba89db7dfa950d506eb26c21b6c0e921ad99b", "compiled_ids_sha256": "5be6ca73d195d1470fed5f5b6987f664dac9260eba28bada034ea920b50c2e7b", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.0, 14.375, 13.6875, 19.75, 14.3125, 14.0625], "runtime_probabilities": [0.00845429114997387, 0.004525256343185902, 0.0022754415404051542, 0.9771832227706909, 0.004251084756106138, 0.0033107481431216], "decide_seconds": 0.16160814000249957, "probabilities_f64_minus_runtime_max_abs": 4.367403594329744e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.019680873473475427, "darwin": 0.005297035235186961, "sagan": 0.0032128142757190123, "ibn_sina": 0.0153274796726719, "einstein": 0.008204208665190147, "socrates": 0.9482775886777565}, "confidence": 0.8452256062905266, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.33269199862843, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "81f07b4c59c6b0e2a3ac361cc52b529a009e086ae13d368060b101bacf0501f0", "compiled_ids_sha256": "7b39dab83d1ca7724c52829cee4e2579aaa9f8c847ef78366aa994a37b94d6d4", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.75, 14.4375, 13.9375, 15.5, 14.875, 19.625], "runtime_probabilities": [0.019680874422192574, 0.005297035910189152, 0.003212814452126622, 0.01532747969031334, 0.008204208686947823, 0.9482776522636414], "decide_seconds": 0.16160104099981254, "probabilities_f64_minus_runtime_max_abs": 6.35858848907489e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.005772566874345978, "einstein": 0.0029026289103701707, "darwin": 0.0012880337711403391, "sagan": 0.0014595334752655058, "hypatia": 0.01778075774786658, "socrates": 0.9707964792210113}, "confidence": 0.9077791451703676, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.92030699757743, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "297cf3e7b0cb93d214dff7d93af8d03747074f169f88e60e5870669375322034", "compiled_ids_sha256": "a6cadf0dd1e21e1c9c35a4acfcef547b396a9320c21c11ec55a31f3ccbce5f64", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.625, 13.9375, 13.125, 13.25, 15.75, 19.75], "runtime_probabilities": [0.0057725668884813786, 0.0029026290867477655, 0.001288033789023757, 0.0014595334650948644, 0.017780758440494537, 0.970796525478363], "decide_seconds": 0.16198991700002807, "probabilities_f64_minus_runtime_max_abs": 4.625735172680834e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.009231029817526272, "sagan": 0.0040962446508055255, "socrates": 0.9415906397967672, "ibn_sina": 0.022144089367137385, "darwin": 0.0033959060907084865, "hypatia": 0.019542090277055217}, "confidence": 0.8308816451440137, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.4771309986245, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "57b0a5c48171094738510bfa2a4aae666897260d2743d7f7cecd03c9a8caecee", "compiled_ids_sha256": "20684a84b7109c4f8c28890a046cd6d8e6afc3f5d60cfb0c06b1ec120e333447", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.25, 14.4375, 19.875, 16.125, 14.25, 16.0], "runtime_probabilities": [0.009231030009686947, 0.004096244461834431, 0.9415906071662903, 0.02214408852159977, 0.003395905951038003, 0.019542088732123375], "decide_seconds": 0.16165950000140583, "probabilities_f64_minus_runtime_max_abs": 3.263047687607212e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9817755724019439, "ibn_sina": 0.0051518853800460365, "hypatia": 0.007979396171705526, "einstein": 0.0018952727145906594, "sagan": 0.0018952727145906594, "darwin": 0.0013026006171232224}, "confidence": 0.935169077977925, "choice": "socrates"}}, "usage": {"input_tokens": 315, "output_tokens": 1}, "latency_ms": 165.3350220003631, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 315, "compiled_prompt_sha256": "513bcaa62245dda4f0599b7a24d1bf790b2259b02178ab05dab7f744b736f641", "compiled_ids_sha256": "d9b510cab54a1f427220ab3ebe3f09ad1bd8c19efbbc6dbf8a6bd01f5fd9a6d7", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.625, 14.375, 14.8125, 13.375, 13.375, 13.0], "runtime_probabilities": [0.9817755222320557, 0.005151885095983744, 0.007979395799338818, 0.0018952726386487484, 0.0018952726386487484, 0.0013026004889979959], "decide_seconds": 0.1616051500022877, "probabilities_f64_minus_runtime_max_abs": 5.016988824380064e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:42Z"} {"id": "c-socrates-01-starter-03", "uid": "c105", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "prompt_sha256_by_order": ["b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f"], "request_sha256": "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "compiled_prompt_sha256_by_order": ["1b43313b82313432c0e61c8d24490e5ee29de32faa2f8898725bc668a978fa6f", "ae6b89e50a5148483737380bffe9acace8cc964e5580e839cc8895265059d463", "87525d5ed757dbe3f1a1f394ce48f41cfeff3a70beab9ab2ee7aec44eca6793f", "93158928479c6a94dccd8832caae1cd3689d9b803d2128164772c4d869e1d61c", "4baed2eca5b55d49908e1c2c5bcfdc217d52427193c34234525c058ceca51e6b", "53075d3ffd6a2d6a6792062761c6c2a82284c8b143a93e98e928e589e597923a"], "compiled_tokens_by_order": [335, 335, 335, 335, 335, 335], "choice": "socrates", "probs": {"darwin": 0.001796451863394044, "einstein": 0.0016876103471879747, "hypatia": 0.0009033127252192645, "ibn_sina": 0.000961571390665898, "sagan": 0.0040483667677124685, "socrates": 0.9906026869058203}, "confidence": 0.9627179463821941, "seconds": 0.17080659800012654, "seconds_all_orders": 1.0248395880007592, "seconds_by_order": [0.17096025900173117, 0.17077642000003834, 0.17062714100029552, 0.1710702679993119, 0.17083626900057425, 0.17056923099880805], "argmax_stable": true, "spread": {"darwin": 0.0017834154395199213, "einstein": 0.0021614298522259485, "hypatia": 0.0015073611968656336, "ibn_sina": 0.0010081104545095441, "sagan": 0.0030154925241978915, "socrates": 0.005268010164793613}, "spread_2dp": {"darwin": 0.0, "einstein": 0.0, "hypatia": 0.0, "ibn_sina": 0.0, "sagan": 0.01, "socrates": 0.0}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.001796451863394044, "einstein": 0.0016876103471879747, "hypatia": 0.0009033127252192645, "ibn_sina": 0.000961571390665898, "sagan": 0.0040483667677124685, "socrates": 0.9906026869058203}, {"ibn_sina": 0.0019112743561856772, "hypatia": 0.0013135983738345337, "sagan": 0.0021657575800980945, "socrates": 0.9900646279639094, "darwin": 0.0015845020941094242, "einstein": 0.002960239631862823}, {"hypatia": 0.001585102011635131, "darwin": 0.001234478687909499, "sagan": 0.0038024620541290627, "ibn_sina": 0.0009031639016761332, "einstein": 0.002035311270982376, "socrates": 0.9904394820736677}, {"ibn_sina": 0.001908368353330725, "einstein": 0.001908368353330725, "darwin": 0.0010214759709442369, "sagan": 0.0043005744710475765, "hypatia": 0.002301931634825227, "socrates": 0.9885592812165215}, {"einstein": 0.0027733233323396366, "sagan": 0.005181250104295986, "socrates": 0.9873714983265734, "ibn_sina": 0.0015801920021926545, "darwin": 0.0022991657966386584, "hypatia": 0.000794570437959593}, {"socrates": 0.9926395084913671, "ibn_sina": 0.0010256920601524806, "hypatia": 0.0009635485197313057, "einstein": 0.0007988097796368744, "sagan": 0.004056690791993528, "darwin": 0.0005157503571187371}], "prompt_tokens": 335, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.001796451863394044, "einstein": 0.0016876103471879747, "hypatia": 0.0009033127252192645, "ibn_sina": 0.000961571390665898, "sagan": 0.0040483667677124685, "socrates": 0.9906026869058203}, "confidence": 0.9627179463821941, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 170.93384899999364, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "1b43313b82313432c0e61c8d24490e5ee29de32faa2f8898725bc668a978fa6f", "compiled_ids_sha256": "8635156608b5843ff3e2d0a3a4e35ad222d413b6cce00d96311035874e998319", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.3125, 13.25, 12.625, 12.6875, 14.125, 19.625], "runtime_probabilities": [0.0017964518629014492, 0.0016876105219125748, 0.0009033127571456134, 0.0009615713497623801, 0.004048367030918598, 0.9906027317047119], "decide_seconds": 0.1669188190026034, "probabilities_f64_minus_runtime_max_abs": 4.4798891596364854e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.0019112743561856772, "hypatia": 0.0013135983738345337, "sagan": 0.0021657575800980945, "socrates": 0.9900646279639094, "darwin": 0.0015845020941094242, "einstein": 0.002960239631862823}, "confidence": 0.9602039209731426, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 170.7524600024044, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "ae6b89e50a5148483737380bffe9acace8cc964e5580e839cc8895265059d463", "compiled_ids_sha256": "fe51e5ec80c56190dd90407e8d0fd3470d1968081556cf8f1521d4d83d86fc5a", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.75, 13.375, 13.875, 20.0, 13.5625, 14.1875], "runtime_probabilities": [0.0019112745067104697, 0.0013135983608663082, 0.0021657575853168964, 0.9900646805763245, 0.0015845021698623896, 0.0029602397698909044], "decide_seconds": 0.16687670000101207, "probabilities_f64_minus_runtime_max_abs": 5.261241509124659e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.001585102011635131, "darwin": 0.001234478687909499, "sagan": 0.0038024620541290627, "ibn_sina": 0.0009031639016761332, "einstein": 0.002035311270982376, "socrates": 0.9904394820736677}, "confidence": 0.9619742406035388, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 170.60349099847372, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "87525d5ed757dbe3f1a1f394ce48f41cfeff3a70beab9ab2ee7aec44eca6793f", "compiled_ids_sha256": "1062e99eefbe42425eac67a4c9a9f9e58fc4bd82ae7f4a114703f46f0d3d2542", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5625, 13.3125, 14.4375, 13.0, 13.8125, 20.0], "runtime_probabilities": [0.0015851020580157638, 0.001234478666447103, 0.003802462015300989, 0.0009031638619489968, 0.0020353111904114485, 0.9904394149780273], "decide_seconds": 0.1667166110019025, "probabilities_f64_minus_runtime_max_abs": 6.709564037699067e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.001908368353330725, "einstein": 0.001908368353330725, "darwin": 0.0010214759709442369, "sagan": 0.0043005744710475765, "hypatia": 0.002301931634825227, "socrates": 0.9885592812165215}, "confidence": 0.9555053261352884, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 171.04674800066277, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "93158928479c6a94dccd8832caae1cd3689d9b803d2128164772c4d869e1d61c", "compiled_ids_sha256": "c7fed4f81f87faf699d6216673c4908d2799b50406ff5284a7234adb84129bd2", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [13.5, 13.5, 12.875, 14.3125, 13.6875, 19.75], "runtime_probabilities": [0.001908368431031704, 0.001908368431031704, 0.0010214759968221188, 0.0043005747720599174, 0.0023019316140562296, 0.9885593056678772], "decide_seconds": 0.16720035699836444, "probabilities_f64_minus_runtime_max_abs": 2.4451355717935996e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.0027733233323396366, "sagan": 0.005181250104295986, "socrates": 0.9873714983265734, "ibn_sina": 0.0015801920021926545, "darwin": 0.0022991657966386584, "hypatia": 0.000794570437959593}, "confidence": 0.9520157414655906, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 170.8120000002964, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "4baed2eca5b55d49908e1c2c5bcfdc217d52427193c34234525c058ceca51e6b", "compiled_ids_sha256": "5536998ca5c3a74592ec468ce00ea29562c01a1872ec9228f1a5e095d2782f86", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 14.625, 19.875, 13.4375, 13.8125, 12.75], "runtime_probabilities": [0.0027733235619962215, 0.005181250628083944, 0.987371563911438, 0.001580192125402391, 0.0022991658188402653, 0.0007945704855956137], "decide_seconds": 0.16676563099827035, "probabilities_f64_minus_runtime_max_abs": 6.558486453922541e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9926395084913671, "ibn_sina": 0.0010256920601524806, "hypatia": 0.0009635485197313057, "einstein": 0.0007988097796368744, "sagan": 0.004056690791993528, "darwin": 0.0005157503571187371}, "confidence": 0.9704047075633891, "choice": "socrates"}}, "usage": {"input_tokens": 335, "output_tokens": 1}, "latency_ms": 170.54523100159713, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 335, "compiled_prompt_sha256": "53075d3ffd6a2d6a6792062761c6c2a82284c8b143a93e98e928e589e597923a", "compiled_ids_sha256": "0959a9a7eba61048e8fe4d8c7762a8dbe3dfa2b38be6b720445cb7768b90724d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.75, 12.875, 12.8125, 12.625, 14.25, 12.1875], "runtime_probabilities": [0.9926395416259766, 0.001025692094117403, 0.0009635484893806279, 0.0007988098077476025, 0.004056690726429224, 0.0005157503765076399], "decide_seconds": 0.16668296200077748, "probabilities_f64_minus_runtime_max_abs": 3.313460950060687e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:43Z"} {"id": "c-socrates-01-starter-03", "uid": "c106", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "prompt_sha256_by_order": ["4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269"], "request_sha256": "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "compiled_prompt_sha256_by_order": ["e89581a53f340a2aeeb9384cd75a6f0aefe0ea16e3efc03c033b3b5f3f2847e8", "437efd1d26cd0c26910a20a82f906c6b9705a9e3fd974ff00769dbe6bdaf0da4", "6851eeebfdc5de3a4d20c0887d25357faaeefc62bba8c472feccf9a3baefe27f", "73cee36163244eb9d3e50cb1af31d65e1916fd5439c18ec841c3a5440bc7c96b", "305844e65da72bfb66ab061ddd0170a9fa893faa80b4c85fabf7f0079e632b50", "1421a4039fffbea0b767ede39fb42de4af16934e20c68dd7afb0522c34e5817c"], "compiled_tokens_by_order": [355, 355, 355, 355, 355, 355], "choice": "socrates", "probs": {"darwin": 0.00615747659037589, "einstein": 0.0039755656870904335, "hypatia": 0.0074273345234129595, "ibn_sina": 0.0030961736702578378, "sagan": 0.006554599711371569, "socrates": 0.9727888498174913}, "confidence": 0.906566521824616, "seconds": 0.17081536633365127, "seconds_all_orders": 1.0248921980019077, "seconds_by_order": [0.17061472100249375, 0.17085248899820726, 0.17058856199946604, 0.17107170799863525, 0.1711324970019632, 0.17063222100114217], "argmax_stable": true, "spread": {"darwin": 0.016692511735565286, "einstein": 0.004884586367960523, "hypatia": 0.0019574404604917953, "ibn_sina": 0.004419824423922754, "sagan": 0.004684086745272622, "socrates": 0.02569642642467418}, "spread_2dp": {"darwin": 0.02, "einstein": 0.01, "hypatia": 0.0, "ibn_sina": 0.01, "sagan": 0.01, "socrates": 0.030000000000000027}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.00615747659037589, "einstein": 0.0039755656870904335, "hypatia": 0.0074273345234129595, "ibn_sina": 0.0030961736702578378, "sagan": 0.006554599711371569, "socrates": 0.9727888498174913}, {"ibn_sina": 0.00684707520077249, "hypatia": 0.007288673611016205, "sagan": 0.0036649752531424354, "socrates": 0.954627854232887, "darwin": 0.019812669030394324, "einstein": 0.007758752671787576}, {"hypatia": 0.005469894062921164, "darwin": 0.006597951679823738, "sagan": 0.002280191086496799, "ibn_sina": 0.0024272507768497357, "einstein": 0.004001859985115569, "socrates": 0.9792228524087929}, {"ibn_sina": 0.004006361270624152, "einstein": 0.0029311165207951567, "darwin": 0.0031201572948290385, "sagan": 0.0025867012507162614, "hypatia": 0.0070313830054742985, "socrates": 0.9803242806575612}, {"einstein": 0.007787729290514187, "sagan": 0.005697624462945722, "socrates": 0.9581931035104536, "ibn_sina": 0.006456254346342651, "darwin": 0.01454939376457964, "hypatia": 0.0073158946251641495}, {"socrates": 0.9709683927633277, "ibn_sina": 0.004786473488296761, "hypatia": 0.00696427783176942, "einstein": 0.0029031429225536637, "sagan": 0.00696427783176942, "darwin": 0.007413435162283032}], "prompt_tokens": 355, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.00615747659037589, "einstein": 0.0039755656870904335, "hypatia": 0.0074273345234129595, "ibn_sina": 0.0030961736702578378, "sagan": 0.006554599711371569, "socrates": 0.9727888498174913}, "confidence": 0.906566521824616, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.58475099838688, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "e89581a53f340a2aeeb9384cd75a6f0aefe0ea16e3efc03c033b3b5f3f2847e8", "compiled_ids_sha256": "67d6cefb26c2e1a4ee433d8b03c2659df0da773a4de7d34f23c8e3cc19bf6703", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.125, 14.75, 13.875, 14.625, 19.625], "runtime_probabilities": [0.006157476920634508, 0.0039755660109221935, 0.007427334785461426, 0.003096173983067274, 0.006554600317031145, 0.97278892993927], "decide_seconds": 0.16641651399913826, "probabilities_f64_minus_runtime_max_abs": 8.012177876715043e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.00684707520077249, "hypatia": 0.007288673611016205, "sagan": 0.0036649752531424354, "socrates": 0.954627854232887, "darwin": 0.019812669030394324, "einstein": 0.007758752671787576}, "confidence": 0.8603200192851879, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.82619899883866, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "437efd1d26cd0c26910a20a82f906c6b9705a9e3fd974ff00769dbe6bdaf0da4", "compiled_ids_sha256": "9c355caacf02925357ec3acdff90bede1d61a96996c01772c656b4544891bf14", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5625, 14.625, 13.9375, 19.5, 15.625, 14.6875], "runtime_probabilities": [0.006847074721008539, 0.007288672961294651, 0.0036649752873927355, 0.9546278715133667, 0.01981266774237156, 0.007758753374218941], "decide_seconds": 0.16672271100105718, "probabilities_f64_minus_runtime_max_abs": 1.728047971383262e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.005469894062921164, "darwin": 0.006597951679823738, "sagan": 0.002280191086496799, "ibn_sina": 0.0024272507768497357, "einstein": 0.004001859985115569, "socrates": 0.9792228524087929}, "confidence": 0.9259062062557645, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.5622219997167, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "6851eeebfdc5de3a4d20c0887d25357faaeefc62bba8c472feccf9a3baefe27f", "compiled_ids_sha256": "2dbadb8722bbc6830cf4d7deec36aef1e8c4e463d32a1d87397cba3151d349b6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.4375, 14.625, 13.5625, 13.625, 14.125, 19.625], "runtime_probabilities": [0.005469894502311945, 0.006597951985895634, 0.0022801910527050495, 0.0024272510781884193, 0.0040018605068326, 0.9792228937149048], "decide_seconds": 0.1665085629974783, "probabilities_f64_minus_runtime_max_abs": 4.130611186425881e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.004006361270624152, "einstein": 0.0029311165207951567, "darwin": 0.0031201572948290385, "sagan": 0.0025867012507162614, "hypatia": 0.0070313830054742985, "socrates": 0.9803242806575612}, "confidence": 0.9291417938182238, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.04596799981664, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "73cee36163244eb9d3e50cb1af31d65e1916fd5439c18ec841c3a5440bc7c96b", "compiled_ids_sha256": "430b0f5076d8b2a4d0f731595846a007bd4023e82a914e182af6e9f23bb6f11d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.0, 13.6875, 13.75, 13.5625, 14.5625, 19.5], "runtime_probabilities": [0.004006361588835716, 0.002931116847321391, 0.0031201576348394156, 0.0025867014192044735, 0.007031382992863655, 0.9803243279457092], "decide_seconds": 0.16697891899821116, "probabilities_f64_minus_runtime_max_abs": 4.728814806398418e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.007787729290514187, "sagan": 0.005697624462945722, "socrates": 0.9581931035104536, "ibn_sina": 0.006456254346342651, "darwin": 0.01454939376457964, "hypatia": 0.0073158946251641495}, "confidence": 0.8670264770517924, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 171.10581699671457, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "305844e65da72bfb66ab061ddd0170a9fa893faa80b4c85fabf7f0079e632b50", "compiled_ids_sha256": "d32001222704ea3f186946ce10cb8bea410d7009ed7f76e50a1bac7f479b0b17", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.8125, 14.5, 19.625, 14.625, 15.4375, 14.75], "runtime_probabilities": [0.007787730079144239, 0.005697624292224646, 0.9581931233406067, 0.0064562545157969, 0.014549393206834793, 0.0073158941231667995], "decide_seconds": 0.16676523099886253, "probabilities_f64_minus_runtime_max_abs": 1.9830153075162116e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.9709683927633277, "ibn_sina": 0.004786473488296761, "hypatia": 0.00696427783176942, "einstein": 0.0029031429225536637, "sagan": 0.00696427783176942, "darwin": 0.007413435162283032}, "confidence": 0.9013948221701775, "choice": "socrates"}}, "usage": {"input_tokens": 355, "output_tokens": 1}, "latency_ms": 170.60573100025067, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 355, "compiled_prompt_sha256": "1421a4039fffbea0b767ede39fb42de4af16934e20c68dd7afb0522c34e5817c", "compiled_ids_sha256": "6c2ed8eba018f460b59ea195da78e10ed5ae741cc48d9199f842e9d0bfe26c6f", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.25, 13.9375, 14.3125, 13.4375, 14.3125, 14.375], "runtime_probabilities": [0.9709684252738953, 0.004786473698914051, 0.006964277476072311, 0.0029031431768089533, 0.006964277476072311, 0.007413435261696577], "decide_seconds": 0.16652940199855948, "probabilities_f64_minus_runtime_max_abs": 3.251056757136439e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:44Z"} {"id": "c-socrates-01-starter-03", "uid": "c107", "task": "kperm", "label": "socrates", "mode": "apus-six-kev-orders", "n_options": 6, "n_perm": 6, "prompt_sha256": "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "prompt_sha256_by_order": ["0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296"], "request_sha256": "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "compiled_prompt_sha256_by_order": ["8501d29a2806beb81cce7eb55e95f965f1f81dc43c69104f947a1d2360bdb71d", "77c3f008f54260781cc6a5a98c14bdcded59801edeea2bf7da2cfd556c6b7e46", "f7a4bd321ebb69865fae5b8e39d55a92b75a7c098cc7a6436e526e0b8b977f30", "1ba1a9e9577f8ba56c3d1708f86214fc84474ba1da0d1f66f9939b6f8302490d", "f3ac6ac3c0fe3d8ed813c3e686bf2b9748592aaf359591a7c06bbb432feef8bd", "4b1c426f4500d6c29cc62fe4b80b73d9f8664fa34b037279b6feb205538d4de8"], "compiled_tokens_by_order": [328, 328, 328, 328, 328, 328], "choice": "socrates", "probs": {"darwin": 0.006783070512892222, "einstein": 0.00870963494173877, "hypatia": 0.005282660627069019, "ibn_sina": 0.004962600399479064, "sagan": 0.028557793538789327, "socrates": 0.9457042399800317}, "confidence": 0.8417460812441853, "seconds": 0.17090027733320312, "seconds_all_orders": 1.0254016639992187, "seconds_by_order": [0.17095673899893882, 0.170849788999476, 0.17069486100081122, 0.17085058000157005, 0.17117007599881617, 0.17087961899960646], "argmax_stable": true, "spread": {"darwin": 0.015456657312002391, "einstein": 0.037123553290319744, "hypatia": 0.005288859680843899, "ibn_sina": 0.019335026851984405, "sagan": 0.05452315427411375, "socrates": 0.08222490734017229}, "spread_2dp": {"darwin": 0.02, "einstein": 0.03, "hypatia": 0.01, "ibn_sina": 0.02, "sagan": 0.049999999999999996, "socrates": 0.07999999999999996}, "choices_by_order": ["socrates", "socrates", "socrates", "socrates", "socrates", "socrates"], "orders": [["darwin", "einstein", "hypatia", "ibn_sina", "sagan", "socrates"], ["ibn_sina", "hypatia", "sagan", "socrates", "darwin", "einstein"], ["hypatia", "darwin", "sagan", "ibn_sina", "einstein", "socrates"], ["ibn_sina", "einstein", "darwin", "sagan", "hypatia", "socrates"], ["einstein", "sagan", "socrates", "ibn_sina", "darwin", "hypatia"], ["socrates", "ibn_sina", "hypatia", "einstein", "sagan", "darwin"]], "probabilities_by_order": [{"darwin": 0.006783070512892222, "einstein": 0.00870963494173877, "hypatia": 0.005282660627069019, "ibn_sina": 0.004962600399479064, "sagan": 0.028557793538789327, "socrates": 0.9457042399800317}, {"ibn_sina": 0.009325406635664831, "hypatia": 0.00565614503881836, "sagan": 0.030576831193042654, "socrates": 0.8935859919163951, "darwin": 0.01636659834397203, "einstein": 0.04448902687210708}, {"hypatia": 0.003928421043777257, "darwin": 0.002874094232710321, "sagan": 0.019950168214777266, "ibn_sina": 0.004181782433397204, "einstein": 0.007812598126055408, "socrates": 0.9612529359492825}, {"ibn_sina": 0.005736236593187722, "einstein": 0.007365473581787335, "darwin": 0.005388695587029216, "sagan": 0.009457455284957035, "hypatia": 0.007365473581787335, "socrates": 0.9646866653712514}, {"einstein": 0.030196180841828263, "sagan": 0.03877266368780694, "socrates": 0.8824617580310791, "ibn_sina": 0.023516809285381607, "darwin": 0.01831490948679608, "hypatia": 0.00673767866710809}, {"socrates": 0.8832250711279862, "ibn_sina": 0.013411064320449617, "hypatia": 0.009217280724621156, "einstein": 0.011835222723159568, "sagan": 0.06398060955907078, "darwin": 0.01833075154471271}], "prompt_tokens": 328, "eval_tokens": 0, "apus_responses": [{"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"darwin": 0.006783070512892222, "einstein": 0.00870963494173877, "hypatia": 0.005282660627069019, "ibn_sina": 0.004962600399479064, "sagan": 0.028557793538789327, "socrates": 0.9457042399800317}, "confidence": 0.8417460812441853, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 170.93104900050093, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "8501d29a2806beb81cce7eb55e95f965f1f81dc43c69104f947a1d2360bdb71d", "compiled_ids_sha256": "da1348b4fbd3b223210453e72685938201ef08da043c35159863b5490ea36eae", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.6875, 14.9375, 14.4375, 14.375, 16.125, 19.625], "runtime_probabilities": [0.006783070508390665, 0.008709635585546494, 0.005282661411911249, 0.004962600767612457, 0.02855779603123665, 0.9457042813301086], "decide_seconds": 0.16697199999907752, "probabilities_f64_minus_runtime_max_abs": 4.135007691807857e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.009325406635664831, "hypatia": 0.00565614503881836, "sagan": 0.030576831193042654, "socrates": 0.8935859919163951, "darwin": 0.01636659834397203, "einstein": 0.04448902687210708}, "confidence": 0.7288562460487582, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 170.8253290016728, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "77c3f008f54260781cc6a5a98c14bdcded59801edeea2bf7da2cfd556c6b7e46", "compiled_ids_sha256": "b0649dc602a40b6a3aef2c126bbaff7cfd5ed4d4f14f93fc364d3a925a16cffb", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [15.1875, 14.6875, 16.375, 19.75, 15.75, 16.75], "runtime_probabilities": [0.009325406514108181, 0.005656144581735134, 0.030576830729842186, 0.8935859799385071, 0.01636659912765026, 0.04448902606964111], "decide_seconds": 0.1669393099982699, "probabilities_f64_minus_runtime_max_abs": 1.1977888014769178e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"hypatia": 0.003928421043777257, "darwin": 0.002874094232710321, "sagan": 0.019950168214777266, "ibn_sina": 0.004181782433397204, "einstein": 0.007812598126055408, "socrates": 0.9612529359492825}, "confidence": 0.8797421370848735, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 170.67129100178136, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "f7a4bd321ebb69865fae5b8e39d55a92b75a7c098cc7a6436e526e0b8b977f30", "compiled_ids_sha256": "cbca5d517839ffd7217054ee63cbe830dcbc5074db77f7af2b4710b3ac8580a6", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.375, 14.0625, 16.0, 14.4375, 15.0625, 19.875], "runtime_probabilities": [0.003928421530872583, 0.0028740945272147655, 0.019950170069932938, 0.004181782715022564, 0.007812599651515484, 0.9612529873847961], "decide_seconds": 0.16684488000100828, "probabilities_f64_minus_runtime_max_abs": 5.143551362607468e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"ibn_sina": 0.005736236593187722, "einstein": 0.007365473581787335, "darwin": 0.005388695587029216, "sagan": 0.009457455284957035, "hypatia": 0.007365473581787335, "socrates": 0.9646866653712514}, "confidence": 0.8834340757972374, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 170.82720000325935, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "1ba1a9e9577f8ba56c3d1708f86214fc84474ba1da0d1f66f9939b6f8302490d", "compiled_ids_sha256": "09b6aaff162d3e86d4434e1e1271911d9cda6a72c822905bfb17f7460e720e88", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [14.5, 14.75, 14.4375, 15.0, 14.75, 19.625], "runtime_probabilities": [0.005736235994845629, 0.007365472614765167, 0.0053886957466602325, 0.009457455016672611, 0.007365472614765167, 0.9646866321563721], "decide_seconds": 0.16702922799959197, "probabilities_f64_minus_runtime_max_abs": 3.3214879291421084e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"einstein": 0.030196180841828263, "sagan": 0.03877266368780694, "socrates": 0.8824617580310791, "ibn_sina": 0.023516809285381607, "darwin": 0.01831490948679608, "hypatia": 0.00673767866710809}, "confidence": 0.7001933519517307, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 171.1452069976076, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "f3ac6ac3c0fe3d8ed813c3e686bf2b9748592aaf359591a7c06bbb432feef8bd", "compiled_ids_sha256": "ee35be70936ae84929a015fe734388ecdf428920a8ec620d93275f85402d2351", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [16.375, 16.625, 19.75, 16.125, 15.875, 14.875], "runtime_probabilities": [0.030196182429790497, 0.038772668689489365, 0.8824618458747864, 0.023516811430454254, 0.018314911052584648, 0.006737678777426481], "decide_seconds": 0.16724989700014703, "probabilities_f64_minus_runtime_max_abs": 8.784370730818836e-08}}}, {"model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "answers": {"q": {"type": "choice", "probabilities": {"socrates": 0.8832250711279862, "ibn_sina": 0.013411064320449617, "hypatia": 0.009217280724621156, "einstein": 0.011835222723159568, "sagan": 0.06398060955907078, "darwin": 0.01833075154471271}, "confidence": 0.714019620692139, "choice": "socrates"}}, "usage": {"input_tokens": 328, "output_tokens": 1}, "latency_ms": 170.85600900099962, "apus": {"q": {"effort": "high", "executed_layers": 32, "projection": "full_head", "calibrated": false, "primitive": "choice", "compiled_tokens": 328, "compiled_prompt_sha256": "4b1c426f4500d6c29cc62fe4b80b73d9f8664fa34b037279b6feb205538d4de8", "compiled_ids_sha256": "c50f8660ccb0895cb05200668ded6977ef98c5a3ec2f592ea50243ce1d18e38d", "candidate_token_ids": [32, 33, 34, 35, 36, 37], "logits": [19.5, 15.3125, 14.9375, 15.1875, 16.875, 15.625], "runtime_probabilities": [0.8832250237464905, 0.013411062769591808, 0.009217280894517899, 0.011835222132503986, 0.06398060917854309, 0.01833074912428856], "decide_seconds": 0.16685592000067118, "probabilities_f64_minus_runtime_max_abs": 4.738149572691697e-08}}}], "orders_source": "Kev-9B's own kperm rows (its server's seeded draw; Kev-4B's are identical)", "prompt_sha256_matches_kev9b": true, "card": "RTX 3090 (fp 33a0b4acb8e8)", "box": "benchbox", "os": {"pretty_name": "Ubuntu 26.04 LTS", "id": "ubuntu", "version_id": "26.04", "kernel": "7.0.0-31-generic", "arch": "x86_64", "source": "/etc/os-release"}, "dry": false, "preflight": false, "effort": "high", "runtime": "openjet_runtime (APUS-OpenJev-v1-4B@422b3741) in-process via deployment/contracts.py (APUS-OpenJev-v1@af276da5), one question per request", "model_revision": "422b3741f8b5c092eeefef847c1ca89d78337d45", "bridge_revision": "af276da53a2e7a6d0fbdbafc9280bf9982b2d3a6", "torch": "2.14.0+cu130", "torch_cuda": "13.0", "transformers": "5.16.1", "tokenizers": "0.23.2", "dtype": "bfloat16", "kernels": "reference (no flash-linear-attention), sdpa", "calibrated": false, "arm": "apus-4b-high", "model": "apus-ailab/APUS-OpenJev-v1-4B@422b3741", "quant": "bf16 (merged LoRA r=8; full weights incl. the vision tower; tied embeddings)", "card_uuid": "GPU-fp:33a0b4acb8e8", "num_ctx": 8191, "power_cap_w": 300, "seed": 20260921, "stamp": "2026-09-28T16:49:45Z"}