{"mode": "kev-choice", "prompt_sha256": "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "n_options": 6, "kev_response": {"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": 193.42580599914072, "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": 0.18685058700066293, "probabilities_f64_minus_runtime_max_abs": 4.2315427606354206e-08}}}, "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": 0.1934930749994237, "prompt_tokens": 279, "eval_tokens": 1, "server_latency_ms": 193.42580599914072, "prompt_eval_ms": null, "total_ms": 193.4930749994237, "done_reason": null, "id": "c-darwin-01-dessert-01", "uid": "c000", "task": "kc", "label": "darwin", "text_sha256": "fcf5cde7b44af0c94fc1aa5d71cb854130f81c4f17c238bcf352c844972bb393", "request_sha256": "ada2e647fb31ad59f271b00730eac9606a376e7312144f898a5078a179b3253d", "compiled_tokens": 279, "compiled_prompt_sha256": "2ca65940a5765f77af59cf411c0b326a074bbc7804d92041b9a96860a2ba21f0", "compiled_ids_sha256": "39875e59687c95458e3c8bef14fa71362355147df3bbbb6895cd7ad2bb135fb8", "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": "2026-09-28T16:44:40Z"} {"mode": "kev-choice", "prompt_sha256": "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "n_options": 6, "kev_response": {"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": 172.77518499759026, "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": 0.16825535899988608, "probabilities_f64_minus_runtime_max_abs": 1.157326046108409e-08}}}, "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": 0.1728038950022892, "prompt_tokens": 379, "eval_tokens": 1, "server_latency_ms": 172.77518499759026, "prompt_eval_ms": null, "total_ms": 172.8038950022892, "done_reason": null, "id": "c-darwin-01-dessert-01", "uid": "c001", "task": "kc", "label": "darwin", "text_sha256": "cca09c399a86e9ba18f164d04a4ceff22429e46d782c68ba7651b4f8f03254c9", "request_sha256": "ae89c5003c7b8c49cffb6cc5b1d1490d93f6203de46ad9f7f01410a034925e66", "compiled_tokens": 379, "compiled_prompt_sha256": "dfdf84665a6deae6826cfe5cb74cbdf7164c3c10210051c62958077957da1655", "compiled_ids_sha256": "935f8d246f67d04de94e49c25e2c59f166760a7b2d6dddf4f9fb2ac6ee141755", "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": "2026-09-28T16:44:40Z"} {"mode": "kev-choice", "prompt_sha256": "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "n_options": 6, "kev_response": {"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": 165.77857900119852, "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": 0.16226044499853742, "probabilities_f64_minus_runtime_max_abs": 1.0869103866939511e-08}}}, "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": 0.16580347800118034, "prompt_tokens": 296, "eval_tokens": 1, "server_latency_ms": 165.77857900119852, "prompt_eval_ms": null, "total_ms": 165.80347800118034, "done_reason": null, "id": "c-darwin-01-dessert-01", "uid": "c002", "task": "kc", "label": "darwin", "text_sha256": "a062080c339f6ce54cbabf2848b449ef14a2780b84a5f4583d04f0156553982d", "request_sha256": "6be4957f07feb5d82a63ae1dca58d96723d78d0c8f0e1a49f0cfafc844c756bc", "compiled_tokens": 296, "compiled_prompt_sha256": "36256d9ffe4d50c3d22a8cefe2d22c38863f669314f6c78f6e3165edd736e631", "compiled_ids_sha256": "bedc1945b96a6e27ccb028c42f119ba2eb225cdab520d6ff987d956978f024cd", "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": "2026-09-28T16:44:40Z"} {"mode": "kev-choice", "prompt_sha256": "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "n_options": 6, "kev_response": {"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": 177.47627800054033, "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": 0.17287088399825734, "probabilities_f64_minus_runtime_max_abs": 9.01330672320455e-08}}}, "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": 0.17750298800092423, "prompt_tokens": 399, "eval_tokens": 1, "server_latency_ms": 177.47627800054033, "prompt_eval_ms": null, "total_ms": 177.50298800092423, "done_reason": null, "id": "c-darwin-01-dessert-02", "uid": "c003", "task": "kc", "label": "darwin", "text_sha256": "485359cc7a1ff70871cdb0ae47661b859bb9bcda6d35a4782bd26dad348398ce", "request_sha256": "1e254c48e8d558d49c1284459198465a15cef2148c47d5875cf4a5f35404bff2", "compiled_tokens": 399, "compiled_prompt_sha256": "604afd02ea457c03d0d85cec823cbc75b5a82d677e33fef5295394b9bc38e733", "compiled_ids_sha256": "77f997cd82656949ba572af316c23fd06bc92bfc51d6b9880262378e26c5a2fd", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "n_options": 6, "kev_response": {"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": 177.20911099968362, "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": 0.172462826998526, "probabilities_f64_minus_runtime_max_abs": 9.826109148569628e-09}}}, "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": 0.17723545099943294, "prompt_tokens": 409, "eval_tokens": 1, "server_latency_ms": 177.20911099968362, "prompt_eval_ms": null, "total_ms": 177.23545099943294, "done_reason": null, "id": "c-darwin-01-dessert-03", "uid": "c004", "task": "kc", "label": "darwin", "text_sha256": "a059154398302a58169460b1f4228781795950865827925fc503638efc0b8cfb", "request_sha256": "7850492fc6179b1b4b4bdd20394ab04399c688ce7f0b9d2de32e671e0627905c", "compiled_tokens": 409, "compiled_prompt_sha256": "f23455f0797804d5728aeff3fdc9d48c4b31d4a7b4b5aa74d195e7140ad387f6", "compiled_ids_sha256": "2fc9c2e9c165a1094386aba821c96f2b186d61c9c48cddddeb2a7985bc7f809a", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "n_options": 6, "kev_response": {"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": 177.50153900124133, "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": 0.17265953500100295, "probabilities_f64_minus_runtime_max_abs": 5.323773977217172e-08}}}, "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": 0.17752837800071575, "prompt_tokens": 411, "eval_tokens": 1, "server_latency_ms": 177.50153900124133, "prompt_eval_ms": null, "total_ms": 177.52837800071575, "done_reason": null, "id": "c-darwin-01-dessert-03", "uid": "c005", "task": "kc", "label": "darwin", "text_sha256": "23d3cdf830a1ee0f835507f044275131684307cb945a80c51538c7b409534789", "request_sha256": "1ef9e9b47aa0b69dc706e2b2d978416f2bd011a8d4c23e06e7b821ffac0d6d06", "compiled_tokens": 411, "compiled_prompt_sha256": "24e4de0b01d855575257e46ab574c911b12ec716af49ef43360d71b0ceb39554", "compiled_ids_sha256": "7edad3a35e2849ccf2f8adce479f5cd518ba5078d299c95eded9c3ee2725b892", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "n_options": 6, "kev_response": {"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": 164.81261499939137, "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": 0.16095180599950254, "probabilities_f64_minus_runtime_max_abs": 1.7075029945523568e-08}}}, "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": 0.16483907499787165, "prompt_tokens": 314, "eval_tokens": 1, "server_latency_ms": 164.81261499939137, "prompt_eval_ms": null, "total_ms": 164.83907499787165, "done_reason": null, "id": "c-darwin-01-main-01", "uid": "c006", "task": "kc", "label": "darwin", "text_sha256": "5dd7557e7f11609ac3d7c3e8fc7175298f308905f62b24085bbccf753b83199d", "request_sha256": "56bc6422a8c3ea07ea10491507d42c62f99c3b12a9c3b03c70a585fa6079a545", "compiled_tokens": 314, "compiled_prompt_sha256": "377f0fd9228b401ffd1752b4418d7a4f949f5c9987b66900bf4451b634fa3b16", "compiled_ids_sha256": "01e1272f80d0db6c2b279d6449fa670109af77f7a80b594d0041b4894a23f3be", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "n_options": 6, "kev_response": {"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": 164.61773799892399, "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": 0.16117056399889407, "probabilities_f64_minus_runtime_max_abs": 2.5151605853590553e-08}}}, "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": 0.16464170700055547, "prompt_tokens": 295, "eval_tokens": 1, "server_latency_ms": 164.61773799892399, "prompt_eval_ms": null, "total_ms": 164.64170700055547, "done_reason": null, "id": "c-darwin-01-main-02", "uid": "c007", "task": "kc", "label": "darwin", "text_sha256": "e5582302f777af6b5f897c4ab2cf415650f9c356ad9b79067cf05fd8ebb156d8", "request_sha256": "7d16a870429eb78dcdb3906b5987ef8ef9229dd0e68f08702c3d72a4601b76cf", "compiled_tokens": 295, "compiled_prompt_sha256": "98b6551eb87406f35159db42e99ab68b6616d9349e25272937a35b9bb812bc44", "compiled_ids_sha256": "1e5073f8ebc6dc8593ad0d9a35127317ad2b076509d4a0a08bc1f851b7949344", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "n_options": 6, "kev_response": {"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": 172.4988959977054, "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": 0.1683077689995116, "probabilities_f64_minus_runtime_max_abs": 1.039352269804894e-07}}}, "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": 0.17252482600088115, "prompt_tokens": 372, "eval_tokens": 1, "server_latency_ms": 172.4988959977054, "prompt_eval_ms": null, "total_ms": 172.52482600088115, "done_reason": null, "id": "c-darwin-01-main-02", "uid": "c008", "task": "kc", "label": "darwin", "text_sha256": "649c28612e897217908c8e2d4d59f51856253ea1c610f2c4ce6cd466b628185e", "request_sha256": "99aa86d7673acbef6b356ef6e1e62bb359780b871a574e286cb0d67fa9e028ce", "compiled_tokens": 372, "compiled_prompt_sha256": "a94c81da04c23301082578b21ecddd1e9b9cf59dce663b78a8e4f45435535a31", "compiled_ids_sha256": "6aa68021cc66abd7749c997fa9d95159d904c84b3398aad7280decc31822aa57", "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": "2026-09-28T16:44:41Z"} {"mode": "kev-choice", "prompt_sha256": "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "n_options": 6, "kev_response": {"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": 177.4365189994569, "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": 0.17277309400014929, "probabilities_f64_minus_runtime_max_abs": 1.051449860511866e-08}}}, "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": 0.17746326899941778, "prompt_tokens": 407, "eval_tokens": 1, "server_latency_ms": 177.4365189994569, "prompt_eval_ms": null, "total_ms": 177.46326899941778, "done_reason": null, "id": "c-darwin-01-main-02", "uid": "c009", "task": "kc", "label": "darwin", "text_sha256": "5d08013bae51d2d1f6280a290baf0b7fbc1b4a3cd4f743a5cb23ab10f96bee6a", "request_sha256": "e1f0f4be015ea8f0987af3df612d4f3a535cc0f500bba24dfada1308176486e1", "compiled_tokens": 407, "compiled_prompt_sha256": "67d1a2d7e7aebb688c009d8c350b3caab3a77e09611db29376cf5a84ebb7f2f5", "compiled_ids_sha256": "b6064067571c6aaaef5c497a87a07eae439d78eab1c8465b4f06fd5fd0319cf5", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "n_options": 6, "kev_response": {"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": 176.586244997452, "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": 0.1719764400004351, "probabilities_f64_minus_runtime_max_abs": 8.339528068113111e-09}}}, "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": 0.176615464999486, "prompt_tokens": 401, "eval_tokens": 1, "server_latency_ms": 176.586244997452, "prompt_eval_ms": null, "total_ms": 176.615464999486, "done_reason": null, "id": "c-darwin-01-main-02", "uid": "c010", "task": "kc", "label": "darwin", "text_sha256": "51ea227691d5922785fa53ce48e8f0d2c1e132ed64a61475fabf1b25d10857d8", "request_sha256": "b336b2cd695a572a1bc843edc2813cd4c22436dec5ab3c120fd4d7ac582be285", "compiled_tokens": 401, "compiled_prompt_sha256": "845aaf3e77ec7fd9ccefeb0369bad6914f74a0a4e47c5038bcac81310111f0a8", "compiled_ids_sha256": "2c82207f9a93c3dc26bdba65b250a344432061dc02ca1018c708ad8067c3121f", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "n_options": 6, "kev_response": {"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": 177.36815900207148, "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": 0.17263279600228998, "probabilities_f64_minus_runtime_max_abs": 5.8774217270851636e-08}}}, "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": 0.17739536900262465, "prompt_tokens": 402, "eval_tokens": 1, "server_latency_ms": 177.36815900207148, "prompt_eval_ms": null, "total_ms": 177.39536900262465, "done_reason": null, "id": "c-darwin-01-main-03", "uid": "c011", "task": "kc", "label": "darwin", "text_sha256": "460ae9db9052213652d36735be593ecd3ddee733eb946607b120386c23188fc5", "request_sha256": "c8f7e1b2d641fcf994218fd2b5f6995d0ac1bf809d856c20e7ffbe37ee02cca5", "compiled_tokens": 402, "compiled_prompt_sha256": "67ed85f77269ae70491a1864dafa1c40fd5c3c6a2590a97dc9bb1b6af2efd7d9", "compiled_ids_sha256": "f12e7fc6a162cdd7569eba4da42ac19c958d6ecfbd2351179f1c50f8bd12014a", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "n_options": 6, "kev_response": {"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": 162.74063200035016, "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": 0.1597774340007163, "probabilities_f64_minus_runtime_max_abs": 1.6856313345137153e-08}}}, "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": 0.16276315200229874, "prompt_tokens": 240, "eval_tokens": 1, "server_latency_ms": 162.74063200035016, "prompt_eval_ms": null, "total_ms": 162.76315200229874, "done_reason": null, "id": "c-darwin-01-starter-01", "uid": "c012", "task": "kc", "label": "darwin", "text_sha256": "6847f4760fc0710fb9f00b2fdee4feda7fb17ee4a2e7a43cc1c094926b2995d6", "request_sha256": "97f03142dc8d0f609da8620d37ed9899a73abb02cf6db98cd2403169e1a8f1d1", "compiled_tokens": 240, "compiled_prompt_sha256": "9e3f59abe3f3521d1414d75abb72215d06b06fabcaa7452231957e31bc3997a0", "compiled_ids_sha256": "79ebf5af4c3e548555cf057c1040dde2bbc5c096909a258efb54835918c41f5f", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "n_options": 6, "kev_response": {"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": 164.2767500015907, "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": 0.1609946749995288, "probabilities_f64_minus_runtime_max_abs": 2.759816375430546e-09}}}, "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": 0.16430175899949973, "prompt_tokens": 269, "eval_tokens": 1, "server_latency_ms": 164.2767500015907, "prompt_eval_ms": null, "total_ms": 164.30175899949973, "done_reason": null, "id": "c-darwin-01-starter-01", "uid": "c013", "task": "kc", "label": "darwin", "text_sha256": "5abf6677c76d7e1ad5b7e964a4153f41b1c6ea65102069c921a1fbb1ee1615de", "request_sha256": "b845ff9ea26f72b9ea56ec38a158e913f20101382e818927c12cf1babe316511", "compiled_tokens": 269, "compiled_prompt_sha256": "4503591e6f0e5386c6e8577056733fa060b0d80509ece075b927a593b418dd73", "compiled_ids_sha256": "fc3048d46a00e63f863e4d4d211b3e8a6c309b8b184a6adb0f036ee0575056f9", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "n_options": 6, "kev_response": {"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": 166.1068460016395, "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": 0.16265260199725162, "probabilities_f64_minus_runtime_max_abs": 6.906878780021941e-08}}}, "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": 0.16613045600024634, "prompt_tokens": 294, "eval_tokens": 1, "server_latency_ms": 166.1068460016395, "prompt_eval_ms": null, "total_ms": 166.13045600024634, "done_reason": null, "id": "c-darwin-01-starter-01", "uid": "c014", "task": "kc", "label": "darwin", "text_sha256": "9dba6878c1f2992e4e0050abb3696a985e529b9b3ccf8a4dbb61b3fdfd630887", "request_sha256": "9a29b19b68b4226e1997d783910f8c1302786371d55a405853d4f64b71916564", "compiled_tokens": 294, "compiled_prompt_sha256": "5d90f70b3e510382c9f8fe500bb3610f93699a121cbbf98eabe2725ae32a0559", "compiled_ids_sha256": "1360cb8f139ab7df74cce524630cc666c0ab3b9f2b63c789ee0517529cc2e1d3", "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": "2026-09-28T16:44:42Z"} {"mode": "kev-choice", "prompt_sha256": "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "n_options": 6, "kev_response": {"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": 164.88442499758094, "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": 0.16171435899741482, "probabilities_f64_minus_runtime_max_abs": 1.4559133898472965e-09}}}, "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": 0.16490865499872598, "prompt_tokens": 273, "eval_tokens": 1, "server_latency_ms": 164.88442499758094, "prompt_eval_ms": null, "total_ms": 164.90865499872598, "done_reason": null, "id": "c-darwin-01-starter-02", "uid": "c015", "task": "kc", "label": "darwin", "text_sha256": "e23a934fa1ab74ba1d2ea29881d37ac5294c3113e530747a58b65988ccce7b3f", "request_sha256": "30a9396b56bc695ae23a32eff428110f7cb586e12fd3fb70164da328567b34bb", "compiled_tokens": 273, "compiled_prompt_sha256": "cc1df5806cd5c60ea952970a2eda3feb0945e82bb626caee1c5af7534e8df1ee", "compiled_ids_sha256": "f791a198d080ca74c8294b34a536763716401dc9291fa42c4a1acd27960ce18f", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "n_options": 6, "kev_response": {"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": 177.04528199828928, "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": 0.17251948600096512, "probabilities_f64_minus_runtime_max_abs": 3.1121107912923662e-09}}}, "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": 0.1770706120005343, "prompt_tokens": 396, "eval_tokens": 1, "server_latency_ms": 177.04528199828928, "prompt_eval_ms": null, "total_ms": 177.0706120005343, "done_reason": null, "id": "c-darwin-01-starter-03", "uid": "c016", "task": "kc", "label": "darwin", "text_sha256": "ed21081e080015da7bfcd7eab39062843beed6a7101b806f582d6b920b521b74", "request_sha256": "e31463776e2d63ca347866822a0a37ee28f8ec2812693c8947769b0b0af4d1d8", "compiled_tokens": 396, "compiled_prompt_sha256": "48c20dba52dff386d2809ed467785e33fb1eac6e82e19292cd9497781a43ff1b", "compiled_ids_sha256": "ec6e841b4247aaa12db66e4d4109704bc2a69c4a9bbb1d5a4325d9879d6dca16", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "n_options": 6, "kev_response": {"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": 171.5712130026077, "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": 0.16718805699929362, "probabilities_f64_minus_runtime_max_abs": 2.278989619242111e-08}}}, "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": 0.17159945299863466, "prompt_tokens": 383, "eval_tokens": 1, "server_latency_ms": 171.5712130026077, "prompt_eval_ms": null, "total_ms": 171.59945299863466, "done_reason": null, "id": "c-darwin-01-starter-03", "uid": "c017", "task": "kc", "label": "darwin", "text_sha256": "62da45e279ad99e725097784fd077fdc2c79b7c185dfcddc965fc9cf95871589", "request_sha256": "d3c89f4f7c26fdb7c98e0288179dc1fda4d9fc4fe1736900021b79f3d0c9b169", "compiled_tokens": 383, "compiled_prompt_sha256": "292b2a1853810df91bacbfc1e9d9b69a191c104ded5e174b75e99205cda89199", "compiled_ids_sha256": "80f60e0dbaddd871d7e686f5baae1c6e910fcda7924288511f08aec576cdfe5e", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "n_options": 6, "kev_response": {"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": 171.57731400220655, "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": 0.16716377799821203, "probabilities_f64_minus_runtime_max_abs": 8.391142869434987e-08}}}, "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": 0.17160446400157525, "prompt_tokens": 373, "eval_tokens": 1, "server_latency_ms": 171.57731400220655, "prompt_eval_ms": null, "total_ms": 171.60446400157525, "done_reason": null, "id": "c-einstein-01-dessert-01", "uid": "c018", "task": "kc", "label": "einstein", "text_sha256": "b781300618d56e5dcd736c459a144267a9267ffc8d98516bbbb276ca5d4718af", "request_sha256": "f3c3ff5751f493093f849d48e30c91c3ca66897d3ab98f8906f24bbf0c764bb3", "compiled_tokens": 373, "compiled_prompt_sha256": "511f397ccbe1d23a7056a582c58d20ec2dc37e7b408ccc37042cd3604b440fdb", "compiled_ids_sha256": "95907c564f0990963bab66233f57176e924cae32017e3e77c3639ed1e429d350", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "n_options": 6, "kev_response": {"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": 171.482804002153, "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": 0.16722953700082144, "probabilities_f64_minus_runtime_max_abs": 1.1697801505228256e-08}}}, "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": 0.17150874399885652, "prompt_tokens": 380, "eval_tokens": 1, "server_latency_ms": 171.482804002153, "prompt_eval_ms": null, "total_ms": 171.50874399885652, "done_reason": null, "id": "c-einstein-01-dessert-01", "uid": "c019", "task": "kc", "label": "einstein", "text_sha256": "eb84fb712ad2783d41fceb39dc694dbb9cf5eb13f70b5718420c281b9e7774de", "request_sha256": "b67cff6debb40976045fe29b9a21746ce0ee5f687719bd311e2c3fe3416d62fd", "compiled_tokens": 380, "compiled_prompt_sha256": "482a99aea3518461af305b2fde5179c25aa0874e22f2f00e33c301df075be35a", "compiled_ids_sha256": "0d8c04b699286490547a95aa4750df3f99b6a6ff4c2170fbcda9228d7de90d1c", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "n_options": 6, "kev_response": {"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": 171.3396359991748, "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": 0.16738661600174964, "probabilities_f64_minus_runtime_max_abs": 7.359364700976201e-09}}}, "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": 0.1713644250012294, "prompt_tokens": 349, "eval_tokens": 1, "server_latency_ms": 171.3396359991748, "prompt_eval_ms": null, "total_ms": 171.3644250012294, "done_reason": null, "id": "c-einstein-01-dessert-02", "uid": "c020", "task": "kc", "label": "einstein", "text_sha256": "4fb0a9de68aead4be7487fe269ed3b70a9895545d5104cbd279be928e1065597", "request_sha256": "ef89dd562e7e2022614d3aeddd490894007adab3496f15b911b8f50b5a8695bf", "compiled_tokens": 349, "compiled_prompt_sha256": "bf3f0bcb078d10695d6d82ba348cba3a8f6b4084459aec08e47402eef4080ca8", "compiled_ids_sha256": "2f7bcc4bbd9219b9afcae08151a8d821af356541e49f69a31dd887caf3fb2636", "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": "2026-09-28T16:44:43Z"} {"mode": "kev-choice", "prompt_sha256": "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "n_options": 6, "kev_response": {"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": 166.58185200139997, "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": 0.16319861900046817, "probabilities_f64_minus_runtime_max_abs": 1.390371337506302e-08}}}, "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": 0.16660556200076826, "prompt_tokens": 287, "eval_tokens": 1, "server_latency_ms": 166.58185200139997, "prompt_eval_ms": null, "total_ms": 166.60556200076826, "done_reason": null, "id": "c-einstein-01-dessert-02", "uid": "c021", "task": "kc", "label": "einstein", "text_sha256": "ddfc77297962cfaed22c1394076f76a103da218260e895d72255351e20d3c1a7", "request_sha256": "dca49448005572a62fd417c9eaf61d7d31ad351962f9ace12cbbf8888be7a516", "compiled_tokens": 287, "compiled_prompt_sha256": "dc15cfd6dc8dccc74630a73d746dd43d7892002655d25aa889cf89abbcc1d5b6", "compiled_ids_sha256": "770cba00074ef91bbf9b863248bc0cf872ece1572ae250a4ff112e7036365292", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "n_options": 6, "kev_response": {"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": 177.03212200285634, "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": 0.1725638159987284, "probabilities_f64_minus_runtime_max_abs": 7.492582965173966e-08}}}, "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": 0.17705879200002528, "prompt_tokens": 407, "eval_tokens": 1, "server_latency_ms": 177.03212200285634, "prompt_eval_ms": null, "total_ms": 177.05879200002528, "done_reason": null, "id": "c-einstein-01-dessert-03", "uid": "c022", "task": "kc", "label": "einstein", "text_sha256": "4fae45c9eb873c9e765d8f804b3b5dbb19d1f7b44fcb0911b23c37f40678fd6e", "request_sha256": "78d241562930238f76945a3734169698bc708243442eeb4f19da867c434d4b2c", "compiled_tokens": 407, "compiled_prompt_sha256": "7f48d90264fa637fb31af7ae8698c998de26360d2468e492bee50203c0985399", "compiled_ids_sha256": "d0e8df03971a580df0512cfa9b64c5331720d7183da1c0497ce954a5948dbfa7", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "n_options": 6, "kev_response": {"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": 165.19128199797706, "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": 0.16169965999870328, "probabilities_f64_minus_runtime_max_abs": 1.64710243222288e-08}}}, "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": 0.16521738200026448, "prompt_tokens": 299, "eval_tokens": 1, "server_latency_ms": 165.19128199797706, "prompt_eval_ms": null, "total_ms": 165.21738200026448, "done_reason": null, "id": "c-einstein-01-main-01", "uid": "c023", "task": "kc", "label": "einstein", "text_sha256": "07aba25149a0c8da9b647404800cdd24da603dcee8ee4d3dea732f27bcb53bf7", "request_sha256": "18ddc18998ab3fb1e833f98202c0e822e1cd8155ccbf0a8a97fcba7933b1f73c", "compiled_tokens": 299, "compiled_prompt_sha256": "398a30a8df523f7677bfca21e4476fae04e1c93179ccaa5354f79989bf208257", "compiled_ids_sha256": "4e8ff629b9c571ecae098a26e5b2d5c09709bf9dbea7851caa35fd43fed66dce", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "n_options": 6, "kev_response": {"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": 171.15613700298127, "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": 0.16707256800145842, "probabilities_f64_minus_runtime_max_abs": 3.613557342241336e-09}}}, "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": 0.17118365699934657, "prompt_tokens": 346, "eval_tokens": 1, "server_latency_ms": 171.15613700298127, "prompt_eval_ms": null, "total_ms": 171.18365699934657, "done_reason": null, "id": "c-einstein-01-main-01", "uid": "c024", "task": "kc", "label": "einstein", "text_sha256": "4f422e6774b951bd581ef4c7876d7e36747ea40e939108fd5e42273d547984fa", "request_sha256": "0cba123d72e8f7f1c3c7f88c01d5c38838fcf30e5686570d971c7add87b763a2", "compiled_tokens": 346, "compiled_prompt_sha256": "31ad7358200e1676880e3e7daab4375eb6f0352b2bc77278e208e561783c83f2", "compiled_ids_sha256": "c9faf3bf2ed711f2c81c34c713922081a5f5829813476075c406186f3052df81", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "n_options": 6, "kev_response": {"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": 170.55421199984266, "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": 0.1662676339983591, "probabilities_f64_minus_runtime_max_abs": 1.2102541635528041e-08}}}, "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": 0.17058055200322997, "prompt_tokens": 384, "eval_tokens": 1, "server_latency_ms": 170.55421199984266, "prompt_eval_ms": null, "total_ms": 170.58055200322997, "done_reason": null, "id": "c-einstein-01-main-02", "uid": "c025", "task": "kc", "label": "einstein", "text_sha256": "17705c749c69c5c0abdfcac80343c256023b8e361313ef21348a7d1a2fa7ce0a", "request_sha256": "af92cf8e77853f7e17b471fab0386d943067a2e1b42c4599e1fd1b6432fd3b11", "compiled_tokens": 384, "compiled_prompt_sha256": "8bc0000acb4ce0e5f74190578a430fa97ca4b3961d56af5f977ae0c41db74aee", "compiled_ids_sha256": "96c01d22640684d10ebbdf561b48b5676cfc3b4bf58f2c5ccd9550d3ab256a3e", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "n_options": 6, "kev_response": {"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": 170.42359200058854, "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": 0.16647878299772856, "probabilities_f64_minus_runtime_max_abs": 1.8539478841006485e-09}}}, "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": 0.17045458199936547, "prompt_tokens": 339, "eval_tokens": 1, "server_latency_ms": 170.42359200058854, "prompt_eval_ms": null, "total_ms": 170.45458199936547, "done_reason": null, "id": "c-einstein-01-main-03", "uid": "c026", "task": "kc", "label": "einstein", "text_sha256": "4f6f62821fd6f35d26a1a816268097bf2622a617d550b39a78052b55c661bfc9", "request_sha256": "fbe5ea900d8ce1bc3acd6791006648e0ced9aec40544e9afe4d501057d4ae8bd", "compiled_tokens": 339, "compiled_prompt_sha256": "babae6f4edd94f8ec21e9cf611f98b3fccee444a918dac8ccf6454902374a632", "compiled_ids_sha256": "6866bdd0a49c4b3aa6a808c4deba4a1c9165cab5449d0a129281ece86c4bd505", "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": "2026-09-28T16:44:44Z"} {"mode": "kev-choice", "prompt_sha256": "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "n_options": 6, "kev_response": {"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": 172.01045099864132, "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": 0.1680591110016394, "probabilities_f64_minus_runtime_max_abs": 3.5384248975312005e-08}}}, "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": 0.17203584099843283, "prompt_tokens": 345, "eval_tokens": 1, "server_latency_ms": 172.01045099864132, "prompt_eval_ms": null, "total_ms": 172.03584099843283, "done_reason": null, "id": "c-einstein-01-main-03", "uid": "c027", "task": "kc", "label": "einstein", "text_sha256": "8ffb6e325b770fdaeee2ad284e09c015327cc5a633bf8c8c97a5cd3a7b6eea00", "request_sha256": "7905abf1a7c38428613208ee4ff2e2dff37c5106352f2c0f16c09b8cf9fff9af", "compiled_tokens": 345, "compiled_prompt_sha256": "5b0777e33fab5b56af64797557c932888e6e56a8386aa298f23774efd807c843", "compiled_ids_sha256": "526e5ba6afca227b1070971298c7ab7274443e599e3f4513ee13c7dcebd97fa0", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "n_options": 6, "kev_response": {"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": 170.5625920003513, "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": 0.16654080299849738, "probabilities_f64_minus_runtime_max_abs": 1.2502768709232726e-08}}}, "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": 0.17058805100168684, "prompt_tokens": 358, "eval_tokens": 1, "server_latency_ms": 170.5625920003513, "prompt_eval_ms": null, "total_ms": 170.58805100168684, "done_reason": null, "id": "c-einstein-01-main-03", "uid": "c028", "task": "kc", "label": "einstein", "text_sha256": "1969e8278b9b5763c7105922f3230596533913fb049d2aa5046a08e38ab07387", "request_sha256": "046c679c54add8987c0def6ffb019370532501b30ece1b193e34c4ac614f29e3", "compiled_tokens": 358, "compiled_prompt_sha256": "23475b0a8910eef4fdc6c4e53fc0b009b8dddcf3b82d3c5b95467ce8d3a7fa19", "compiled_ids_sha256": "74797ebb675858e67a4e4dc5542e122d2f6918fddd42e509d7955ac569f7cd1a", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "n_options": 6, "kev_response": {"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": 176.59743499825709, "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": 0.17213937999986229, "probabilities_f64_minus_runtime_max_abs": 3.8328580531654666e-08}}}, "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": 0.17662400499830255, "prompt_tokens": 402, "eval_tokens": 1, "server_latency_ms": 176.59743499825709, "prompt_eval_ms": null, "total_ms": 176.62400499830255, "done_reason": null, "id": "c-einstein-01-main-03", "uid": "c029", "task": "kc", "label": "einstein", "text_sha256": "09fcaf85056d117752e4cb783fd351b35d15712f09db521a1956d6e0f9c7476e", "request_sha256": "1c86e1f79ad9ee96dfc6e80b1c4bfb37ccb503740730c4dcdc4de0bbe9576979", "compiled_tokens": 402, "compiled_prompt_sha256": "1f30f52d13f815504e7bf1165146b0c24247b90d1788ddb8d1a5b6473d203fe6", "compiled_ids_sha256": "8e661bcb0784f987ceba2e82943447a2e43cffd2e0348e8aba9d9b56cec2f1d5", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "n_options": 6, "kev_response": {"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": 165.68906899919966, "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": 0.16204053700130316, "probabilities_f64_minus_runtime_max_abs": 4.5673535398194076e-08}}}, "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": 0.16571324899996398, "prompt_tokens": 305, "eval_tokens": 1, "server_latency_ms": 165.68906899919966, "prompt_eval_ms": null, "total_ms": 165.71324899996398, "done_reason": null, "id": "c-einstein-01-starter-01", "uid": "c030", "task": "kc", "label": "einstein", "text_sha256": "08e02bf156cc40d29eb210f32660351ca76ed14d072452ab76abe5df36f30cf1", "request_sha256": "08395ee9a394a652ac502ba3783881f851a097c91c2d8c801bc9dd19465066fa", "compiled_tokens": 305, "compiled_prompt_sha256": "53cefe6d23d0ac7f3c46766dc20453dc7cb5499f889b97160ea8df183525bf2a", "compiled_ids_sha256": "f3e5ac56a91af44389f871a8caca89d7e9b0d855b71d7a5f4fe583f155b126f2", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "n_options": 6, "kev_response": {"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": 166.54714200194576, "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": 0.16341196699795546, "probabilities_f64_minus_runtime_max_abs": 4.027589561861333e-08}}}, "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": 0.16657155199936824, "prompt_tokens": 267, "eval_tokens": 1, "server_latency_ms": 166.54714200194576, "prompt_eval_ms": null, "total_ms": 166.57155199936824, "done_reason": null, "id": "c-einstein-01-starter-02", "uid": "c031", "task": "kc", "label": "einstein", "text_sha256": "a36451d05d663fe9e9be29012d01cc938a3f5f796503647444fa24af9f43c037", "request_sha256": "06550a0c070754bfa382273603bae053348544611ea06fc60c3f8f78f6f0e0b7", "compiled_tokens": 267, "compiled_prompt_sha256": "2c70462d4e85f4e64653498d9e754bf182bcbee31c8f1c5ece9f76079ffaacbf", "compiled_ids_sha256": "fc5d36802beb2fda14c96e74d28c496080c202d461eda3e170bc4d68d9dc027a", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "n_options": 6, "kev_response": {"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": 164.71215699857567, "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": 0.1613199919993349, "probabilities_f64_minus_runtime_max_abs": 1.3224480704954189e-08}}}, "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": 0.16473615699942457, "prompt_tokens": 294, "eval_tokens": 1, "server_latency_ms": 164.71215699857567, "prompt_eval_ms": null, "total_ms": 164.73615699942457, "done_reason": null, "id": "c-einstein-01-starter-02", "uid": "c032", "task": "kc", "label": "einstein", "text_sha256": "514fdcb80f9e03597268ab55271c5a422f6c2fad64c0388d69b1a9fc12a70238", "request_sha256": "8d3b9b5c684b9c4f71fe2c80ef2a0abb1bf762cb9eca32cc2797975293e622db", "compiled_tokens": 294, "compiled_prompt_sha256": "370a32f6cf6865231ae43ab9a2c0130f1bd393f6ff9c93f7322c5eec6f69a7f6", "compiled_ids_sha256": "45bd3725a496f6775e0e3d110492d7c95aeb7acbd0c5e55b8758c122228a0ab4", "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": "2026-09-28T16:44:45Z"} {"mode": "kev-choice", "prompt_sha256": "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "n_options": 6, "kev_response": {"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": 177.34397900130716, "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": 0.17305648299952736, "probabilities_f64_minus_runtime_max_abs": 3.756650823749652e-08}}}, "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": 0.17737266900076065, "prompt_tokens": 387, "eval_tokens": 1, "server_latency_ms": 177.34397900130716, "prompt_eval_ms": null, "total_ms": 177.37266900076065, "done_reason": null, "id": "c-einstein-01-starter-03", "uid": "c033", "task": "kc", "label": "einstein", "text_sha256": "b6f364845ed13528ab3c01f57ef05b30b6bf428289727b3932852952d458fa8d", "request_sha256": "ef45da2316173e7abeea5d9f50b805e7e6101970e193e303448a09b24841b95c", "compiled_tokens": 387, "compiled_prompt_sha256": "32dea34ab78631d1128469f41a9a9b72979f369974ba4b1ff7cdc18dca902d44", "compiled_ids_sha256": "4844b40c337845fad28c4f6e260cd19b165cd0fb8800c232d4d2050f7333a62b", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "n_options": 6, "kev_response": {"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": 176.01840000133961, "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": 0.17205764000027557, "probabilities_f64_minus_runtime_max_abs": 3.957137395538268e-09}}}, "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": 0.17604523000045447, "prompt_tokens": 343, "eval_tokens": 1, "server_latency_ms": 176.01840000133961, "prompt_eval_ms": null, "total_ms": 176.04523000045447, "done_reason": null, "id": "c-einstein-01-starter-03", "uid": "c034", "task": "kc", "label": "einstein", "text_sha256": "410e86ba344ca5a2f88341617f6d75ef59d1ee1233f886347c1cef0b365b938c", "request_sha256": "70c61a24e7389825115e809e90590b0de304acdcec9da8139878d698c530832c", "compiled_tokens": 343, "compiled_prompt_sha256": "3048ef259ba2c81faa325128892aa67888378d9abaef64bcc7e191dc46c7930d", "compiled_ids_sha256": "dee894996ae3ae3d47634ad4330ad3215bf4766aff9df1233b15480c06ddd321", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "n_options": 6, "kev_response": {"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": 164.5742379987496, "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": 0.16131795199908083, "probabilities_f64_minus_runtime_max_abs": 6.639331817659411e-08}}}, "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": 0.1645977469997888, "prompt_tokens": 278, "eval_tokens": 1, "server_latency_ms": 164.5742379987496, "prompt_eval_ms": null, "total_ms": 164.5977469997888, "done_reason": null, "id": "c-einstein-01-starter-03", "uid": "c035", "task": "kc", "label": "einstein", "text_sha256": "67ab5560f37aea121e4812ee7208fe88b05d544de1313ee421b13d1f809a395d", "request_sha256": "36e36fd875dcb6fcd30f8ad4a732fc84604c5605fd51588de8cdf3fd37bcb156", "compiled_tokens": 278, "compiled_prompt_sha256": "8f4e67e82afe236dfe30a2d5eeec24f6f2778b73186fa1f61532a001b7adee48", "compiled_ids_sha256": "bc3a2e6b1255920114fcfe262b849bf8a7b90aa700f811ca5a13e3dea051bd85", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "n_options": 6, "kev_response": {"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": 176.8596729998535, "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": 0.17231500900015817, "probabilities_f64_minus_runtime_max_abs": 2.217811367888345e-08}}}, "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": 0.17688905300019542, "prompt_tokens": 395, "eval_tokens": 1, "server_latency_ms": 176.8596729998535, "prompt_eval_ms": null, "total_ms": 176.88905300019542, "done_reason": null, "id": "c-hypatia-01-dessert-01", "uid": "c036", "task": "kc", "label": "hypatia", "text_sha256": "a5de183d7f21379964de4518ebf80f3053096a36e1dd875744e7d22e87edc7c1", "request_sha256": "53410880d12779c0c4c4e4becb3d6c57efa5cca73c859c565c763499a1fba816", "compiled_tokens": 395, "compiled_prompt_sha256": "4231600771fa3f6d01b11d9780088dab4c81fa8caa77568d942fc82a7a6b4f8c", "compiled_ids_sha256": "7ef9a0540d1930f31f588689d49c705e105a918541d1c90201320219a03b21c8", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "n_options": 6, "kev_response": {"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": 179.64825200033374, "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": 0.17601944899797672, "probabilities_f64_minus_runtime_max_abs": 1.4991740182601632e-08}}}, "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": 0.1796757219999563, "prompt_tokens": 323, "eval_tokens": 1, "server_latency_ms": 179.64825200033374, "prompt_eval_ms": null, "total_ms": 179.6757219999563, "done_reason": null, "id": "c-hypatia-01-dessert-02", "uid": "c037", "task": "kc", "label": "hypatia", "text_sha256": "01a0fda2d7f9187d9c9d4637119020d2e87fb932554aecc4f25d5e1049099cfd", "request_sha256": "60ca01400728b39379b62e800f8729bb21f28627f8ef74fdb91406ffc80d039b", "compiled_tokens": 323, "compiled_prompt_sha256": "64cf4ddb66e4162fd4413e7c5eecfae78ed5607c7644b898fafd96f028bf8081", "compiled_ids_sha256": "2edffcb25f65010db4bf8ec46a80e7f11423c9168ec99e3b3ac49296338f2c48", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "n_options": 6, "kev_response": {"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": 164.92786500020884, "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": 0.16166857999996864, "probabilities_f64_minus_runtime_max_abs": 3.7721491374931304e-08}}}, "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": 0.1649527150002541, "prompt_tokens": 274, "eval_tokens": 1, "server_latency_ms": 164.92786500020884, "prompt_eval_ms": null, "total_ms": 164.9527150002541, "done_reason": null, "id": "c-hypatia-01-dessert-02", "uid": "c038", "task": "kc", "label": "hypatia", "text_sha256": "d19c1175d3216a8a0bfd4d0953337a3682e2eacf13a33a217ea2c81dacc6ba81", "request_sha256": "cea561710397d08713919835eaa6badd506abefe21e9793915736cbcf571db70", "compiled_tokens": 274, "compiled_prompt_sha256": "b813667369f4e70abaa4bbd36b3b8b690e78c38b9ec85f4f3d5388a491e9ce79", "compiled_ids_sha256": "1356d08876a1c4cf2b6410ee614660ec7bf937bb468b506992fcb7bf08d22f64", "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": "2026-09-28T16:44:46Z"} {"mode": "kev-choice", "prompt_sha256": "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "n_options": 6, "kev_response": {"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": 164.8717560019577, "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": 0.16141135200086865, "probabilities_f64_minus_runtime_max_abs": 2.1602312688662195e-08}}}, "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": 0.1648957050019817, "prompt_tokens": 298, "eval_tokens": 1, "server_latency_ms": 164.8717560019577, "prompt_eval_ms": null, "total_ms": 164.8957050019817, "done_reason": null, "id": "c-hypatia-01-dessert-03", "uid": "c039", "task": "kc", "label": "hypatia", "text_sha256": "5f26c48e950fedd038bfdac0246a243725b4544c160a6d2b3bdc1bc678cd5d07", "request_sha256": "a92667bd1952495e79468ae067425395d5511079bdf1f066abafa0182c57620e", "compiled_tokens": 298, "compiled_prompt_sha256": "21829cbfc35a4be0d7b256334b6e53727cf63ff22411b3b96a97b55be3b983e1", "compiled_ids_sha256": "90ee18b88d0e89b3999ace8ac624b7a7824208a64fcb7a98548c6bbcf7208086", "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": "2026-09-28T16:44:47Z"} {"mode": "kev-choice", "prompt_sha256": "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "n_options": 6, "kev_response": {"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": 172.61557599704247, "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": 0.16853987699869322, "probabilities_f64_minus_runtime_max_abs": 1.1073773897862793e-08}}}, "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": 0.17264000499926624, "prompt_tokens": 368, "eval_tokens": 1, "server_latency_ms": 172.61557599704247, "prompt_eval_ms": null, "total_ms": 172.64000499926624, "done_reason": null, "id": "c-hypatia-01-main-01", "uid": "c040", "task": "kc", "label": "hypatia", "text_sha256": "6d5bf83d9dbfa6c8fa54c2ac6f17730078b8b12a6895818e9d5bae99813c9910", "request_sha256": "6dae1f58ef7ea31d6acf53cc5f4fc8a8254827e1ec602b97859fd641edded372", "compiled_tokens": 368, "compiled_prompt_sha256": "a85f301fbcbe5359d5137e8ce9e30a12f45037248d0741358fb7ff25b1b3eafd", "compiled_ids_sha256": "c6dbc71f4fcb0e3fc82bd7695103d89d45b3a9f018a7f14f1ed0ba8bed8dbf04", "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": "2026-09-28T16:44:47Z"} {"mode": "kev-choice", "prompt_sha256": "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "n_options": 6, "kev_response": {"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": 165.1044140016893, "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": 0.1616261700000905, "probabilities_f64_minus_runtime_max_abs": 4.60396563928267e-09}}}, "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": 0.16512840399809647, "prompt_tokens": 297, "eval_tokens": 1, "server_latency_ms": 165.1044140016893, "prompt_eval_ms": null, "total_ms": 165.12840399809647, "done_reason": null, "id": "c-hypatia-01-main-01", "uid": "c041", "task": "kc", "label": "hypatia", "text_sha256": "f1b2cfbff4f5841853cb9507da2768ff094aeeca9b559b51245ee33878dffeb2", "request_sha256": "c5e40b39c14beb9a0b05e0922a154ec9943477a8ebdd3b239151420d3872133c", "compiled_tokens": 297, "compiled_prompt_sha256": "937427227f5f99118cb41f68bee040c493d9bec3da24346374497e87fd4b0c16", "compiled_ids_sha256": "f2a92c82350a676ec89ba5633141c479e87276520dc9842cd8d2d1356c426518", "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": "2026-09-28T16:44:47Z"} {"mode": "kev-choice", "prompt_sha256": "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "n_options": 6, "kev_response": {"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": 176.8002630014962, "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": 0.17234795799959102, "probabilities_f64_minus_runtime_max_abs": 3.370759193987993e-08}}}, "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": 0.17682636300014565, "prompt_tokens": 388, "eval_tokens": 1, "server_latency_ms": 176.8002630014962, "prompt_eval_ms": null, "total_ms": 176.82636300014565, "done_reason": null, "id": "c-hypatia-01-main-01", "uid": "c042", "task": "kc", "label": "hypatia", "text_sha256": "e252c70e869af17f3ef75ee21e1cac261c8c996f84df756e6abdf0a420baf24b", "request_sha256": "79fb4c421b342027371a354bfc724fa64f5db5d74a395e1f0134e383b6932a60", "compiled_tokens": 388, "compiled_prompt_sha256": "9e620d83c4565136473c23f396206defac1ffa16bd69dabc4a8fb76aa81893ae", "compiled_ids_sha256": "5cdc4054cab42c152b7033db49be91482544fff96b21837effc480511601a69b", "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": "2026-09-28T16:44:47Z"} {"mode": "kev-choice", "prompt_sha256": "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "n_options": 6, "kev_response": {"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": 170.75181999825872, "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": 0.16689820000101463, "probabilities_f64_minus_runtime_max_abs": 4.740030312699872e-08}}}, "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": 0.17077683999741566, "prompt_tokens": 339, "eval_tokens": 1, "server_latency_ms": 170.75181999825872, "prompt_eval_ms": null, "total_ms": 170.77683999741566, "done_reason": null, "id": "c-hypatia-01-main-01", "uid": "c043", "task": "kc", "label": "hypatia", "text_sha256": "c9971b306d10c18655f624323a0657456d02d8b9f6f8c8a390c53a0144a20353", "request_sha256": "1171383a350272ae6dd50d3be1dda29c6229f3879efb1331a0015d33d4fa3568", "compiled_tokens": 339, "compiled_prompt_sha256": "96b3447c5f6b1e9a91ad93d8847b48d6895167515fc8658b556b546cd722af75", "compiled_ids_sha256": "a5315fe99a572917cd0eab60dbea5014cc09871c3719dbb84d88bd7901126ebf", "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": "2026-09-28T16:44:47Z"} {"mode": "kev-choice", "prompt_sha256": "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "n_options": 6, "kev_response": {"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": 164.837576001446, "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": 0.16160614999898826, "probabilities_f64_minus_runtime_max_abs": 4.186891777724355e-08}}}, "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": 0.16486149599950295, "prompt_tokens": 275, "eval_tokens": 1, "server_latency_ms": 164.837576001446, "prompt_eval_ms": null, "total_ms": 164.86149599950295, "done_reason": null, "id": "c-hypatia-01-main-01", "uid": "c044", "task": "kc", "label": "hypatia", "text_sha256": "241d6cdc11e0749d2f38d145fbd4bda015a93b0c01b2e0692f7d374cdd1fc0a9", "request_sha256": "550dcbbf4c9113c575eb00cf7dd4c978283e0d887b4fa383e793b9d3768433b0", "compiled_tokens": 275, "compiled_prompt_sha256": "12910ec5c352e9fa1330272cb09c7a1e3ca9ef566bd28df789137a8794a156bc", "compiled_ids_sha256": "bf822f6c8b673ada113561218d839770895abe054bccedd183916316a0e84576", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "n_options": 6, "kev_response": {"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": 170.72379000092042, "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": 0.1667382199993881, "probabilities_f64_minus_runtime_max_abs": 2.393869175110197e-08}}}, "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": 0.1707481599987659, "prompt_tokens": 349, "eval_tokens": 1, "server_latency_ms": 170.72379000092042, "prompt_eval_ms": null, "total_ms": 170.7481599987659, "done_reason": null, "id": "c-hypatia-01-main-02", "uid": "c045", "task": "kc", "label": "hypatia", "text_sha256": "990f4efb9ecf643ec7583c478c70e24062ee44b2b11fc92c39414c900af4f50c", "request_sha256": "a7501a3b3149ec22eb93de5cfbdf2becc70db679c219e6f88ab7072e69212da2", "compiled_tokens": 349, "compiled_prompt_sha256": "53ea90fe5d10904b59900dd3288e3f69d354a8b53be8832d2ed7310f00a05a78", "compiled_ids_sha256": "fc14aba23487d6da377ebf5043577838bb63cffc65b91089be20b810df6d2bad", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "n_options": 6, "kev_response": {"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": 172.0297700012452, "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": 0.16796789200088824, "probabilities_f64_minus_runtime_max_abs": 2.1597831412201174e-08}}}, "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": 0.17205557000124827, "prompt_tokens": 364, "eval_tokens": 1, "server_latency_ms": 172.0297700012452, "prompt_eval_ms": null, "total_ms": 172.05557000124827, "done_reason": null, "id": "c-hypatia-01-main-03", "uid": "c046", "task": "kc", "label": "hypatia", "text_sha256": "2f798896840b41edb036a413b36f53af135fdff331b30e07736bc0e6c8953bd7", "request_sha256": "51091ba903351f76f1c9074582ea4abcadfbf07b73645803c2453268f86cf0fe", "compiled_tokens": 364, "compiled_prompt_sha256": "981f394b92cdbfe83d1629e86ff1043844d3ddec1aa960eee15412baa3b51c00", "compiled_ids_sha256": "e1be0ad9a565296219335b76cdbbdeef9cf60282395d86cca26940dc6907053b", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "n_options": 6, "kev_response": {"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": 171.59371399975498, "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": 0.16772864300219226, "probabilities_f64_minus_runtime_max_abs": 2.329998771699593e-08}}}, "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": 0.17162316399844713, "prompt_tokens": 329, "eval_tokens": 1, "server_latency_ms": 171.59371399975498, "prompt_eval_ms": null, "total_ms": 171.62316399844713, "done_reason": null, "id": "c-hypatia-01-main-03", "uid": "c047", "task": "kc", "label": "hypatia", "text_sha256": "76cf8cc11125b2570e9c8e373627b9b8bce42a3ed2f50594fad3be025b99d1a8", "request_sha256": "ef8ed0aee48a818fd4b6f519a6d69a1ee3cdac3aca20b9c6b8fec197d1a29a2f", "compiled_tokens": 329, "compiled_prompt_sha256": "5e16fbf7126da6744631cdedb3c9e498a2b9052c185376919d00b1f279b12f9c", "compiled_ids_sha256": "07c699060a078fa6ae02cca28988321df21cdd9ed54996d09648ee655e350dce", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "n_options": 6, "kev_response": {"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": 177.2362109986716, "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": 0.1724402369982272, "probabilities_f64_minus_runtime_max_abs": 1.0957523977062067e-07}}}, "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": 0.17726516999755404, "prompt_tokens": 407, "eval_tokens": 1, "server_latency_ms": 177.2362109986716, "prompt_eval_ms": null, "total_ms": 177.26516999755404, "done_reason": null, "id": "c-hypatia-01-main-03", "uid": "c048", "task": "kc", "label": "hypatia", "text_sha256": "50041391af6fba1f4bd767736233b10e810a2891780598b23f019220d35ebab1", "request_sha256": "518f70391848f4160de4d2ca4b3dfa9256f61398a04ebe65ce467d3f4df3d9ec", "compiled_tokens": 407, "compiled_prompt_sha256": "1758a9f1601e86873bbd3a7f9387586fd8779e1252ba78aa05c488753ae18af9", "compiled_ids_sha256": "c2e4ce3277e1b394388c8fbe6816aea6f422bb1f7b5f4acc516099b73569fad1", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "n_options": 6, "kev_response": {"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": 164.30958899945836, "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": 0.16113975400003255, "probabilities_f64_minus_runtime_max_abs": 2.929098591453183e-08}}}, "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": 0.16433572900132276, "prompt_tokens": 272, "eval_tokens": 1, "server_latency_ms": 164.30958899945836, "prompt_eval_ms": null, "total_ms": 164.33572900132276, "done_reason": null, "id": "c-hypatia-01-starter-01", "uid": "c049", "task": "kc", "label": "hypatia", "text_sha256": "b3cbd4a18274f64a6bde319e80f25054e12322fb09acc0c3d522792b7b49ede8", "request_sha256": "2c8f8199cffc7ca570c61b59355b5cb2b0d15104ad1e57f3fd7692a8b2ec257e", "compiled_tokens": 272, "compiled_prompt_sha256": "86edd11982c21ff733510b65fb8061cf99f203a5783f0d267f8b7922baef66d7", "compiled_ids_sha256": "0632451188e0cd8274e38703ba69b1d81b731677e7550055093537363c0d1780", "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": "2026-09-28T16:44:48Z"} {"mode": "kev-choice", "prompt_sha256": "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "n_options": 6, "kev_response": {"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": 170.2515940014564, "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": 0.16658400199958123, "probabilities_f64_minus_runtime_max_abs": 3.1893526597492894e-08}}}, "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": 0.17027593399689067, "prompt_tokens": 326, "eval_tokens": 1, "server_latency_ms": 170.2515940014564, "prompt_eval_ms": null, "total_ms": 170.27593399689067, "done_reason": null, "id": "c-hypatia-01-starter-01", "uid": "c050", "task": "kc", "label": "hypatia", "text_sha256": "6588c54ef23e640e7f57252c699affd5dd6b2af55b7d2263da46b57ca15b64fd", "request_sha256": "49cf9584bdd714487fe44d03d0cf094bffb7ddabf39f43dc80799d3de073d3f7", "compiled_tokens": 326, "compiled_prompt_sha256": "ecbc5708033f479cd2caa368a60bec322a499a7331fb8a6eacdc2dac8b96a819", "compiled_ids_sha256": "3213fbf2a47076131e60a1856d79082d20d04b59cb63f46819df660c2b4096ff", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "n_options": 6, "kev_response": {"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": 164.8893150013464, "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": 0.16144264200192993, "probabilities_f64_minus_runtime_max_abs": 2.8795803519443552e-09}}}, "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": 0.16491257500092615, "prompt_tokens": 303, "eval_tokens": 1, "server_latency_ms": 164.8893150013464, "prompt_eval_ms": null, "total_ms": 164.91257500092615, "done_reason": null, "id": "c-hypatia-01-starter-02", "uid": "c051", "task": "kc", "label": "hypatia", "text_sha256": "584240bc99dcaa76ab12a358683ec6fe3c209a90fc88a943d6e23be4b03fd7fe", "request_sha256": "d445b0eba928e2f123e2a696fc3349b65c89382bdce54490773f3593ea646bbd", "compiled_tokens": 303, "compiled_prompt_sha256": "d3cc18c537cd2524953ff31d3b343b481a694d8832a6516cf95f0af6ac3a93fc", "compiled_ids_sha256": "4176637f3b7f7e0c1eaf039089f3e571cafaabcf2dff27018fe9ae1943bf4ef4", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "n_options": 6, "kev_response": {"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": 168.25995900217094, "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": 0.16506352399665047, "probabilities_f64_minus_runtime_max_abs": 4.313461610117031e-08}}}, "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": 0.1682824889976473, "prompt_tokens": 273, "eval_tokens": 1, "server_latency_ms": 168.25995900217094, "prompt_eval_ms": null, "total_ms": 168.2824889976473, "done_reason": null, "id": "c-hypatia-01-starter-02", "uid": "c052", "task": "kc", "label": "hypatia", "text_sha256": "e91537d8f3532b1cdb47129a57ad062155f8ced9d488616dd4276ec4164938fb", "request_sha256": "44eede8e8a6cd0f0ec8f329973c8450624acfaaf39b0e500af444335f60a264d", "compiled_tokens": 273, "compiled_prompt_sha256": "25078b32820a0c397cc2958b1cb9a27364e725d270a1e669793c995eab4ad072", "compiled_ids_sha256": "9403e074d597096fdde8ed17e223082b95d8f4cb2f6a7e59d08436d83892a5ca", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "n_options": 6, "kev_response": {"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": 166.32376400229987, "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": 0.16270361200076877, "probabilities_f64_minus_runtime_max_abs": 4.440437539887654e-08}}}, "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": 0.16634859400073765, "prompt_tokens": 309, "eval_tokens": 1, "server_latency_ms": 166.32376400229987, "prompt_eval_ms": null, "total_ms": 166.34859400073765, "done_reason": null, "id": "c-hypatia-01-starter-02", "uid": "c053", "task": "kc", "label": "hypatia", "text_sha256": "8d7b0004f9f8bdcc93faf53126599249fb43b754daa1ab89d93c77cb67b3310d", "request_sha256": "a66cb47a052715ec47cff3da36e4d6f8643ac70737abfd23a65b0300a9c4f131", "compiled_tokens": 309, "compiled_prompt_sha256": "ae3c350b97daa5c6854c06d3ed1c5bfc8800493c92943b74514bed4d49857d5a", "compiled_ids_sha256": "55a7c5fbd189b1b2d5df695ec9fc001b672fd7fe50d6a7ee300faac082ac3ee1", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "n_options": 6, "kev_response": {"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": 165.17868300070404, "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": 0.1614798709997558, "probabilities_f64_minus_runtime_max_abs": 2.1914870862449476e-08}}}, "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": 0.16520218300138367, "prompt_tokens": 311, "eval_tokens": 1, "server_latency_ms": 165.17868300070404, "prompt_eval_ms": null, "total_ms": 165.20218300138367, "done_reason": null, "id": "c-ibn_sina-01-dessert-01", "uid": "c054", "task": "kc", "label": "ibn_sina", "text_sha256": "b299c905839b0a3ea80642f3563ed7da112315705aac5520df3856a5ae1de6c9", "request_sha256": "2bfa4bdd272a13a8e0a461cd9394d00a24d529a16ca98d344a304f473afee10d", "compiled_tokens": 311, "compiled_prompt_sha256": "066735836834e07c7890804951ba5a2894b840544a1b10249ec2970aea3a1c62", "compiled_ids_sha256": "8b485c39edfb00d587270bcc96d96b07d12859a38e102a757f143839cbda8eb7", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "n_options": 6, "kev_response": {"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": 170.47319199991762, "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": 0.16679025100165745, "probabilities_f64_minus_runtime_max_abs": 8.962433017956073e-10}}}, "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": 0.1704971420003858, "prompt_tokens": 323, "eval_tokens": 1, "server_latency_ms": 170.47319199991762, "prompt_eval_ms": null, "total_ms": 170.4971420003858, "done_reason": null, "id": "c-ibn_sina-01-dessert-02", "uid": "c055", "task": "kc", "label": "ibn_sina", "text_sha256": "08a2d58741f0e2b5e18564f563f0c14b6a31eedffbd7d8bd3650b24b79ed9e26", "request_sha256": "a316eee1d777f5acec746f93cc5f922a09147773e8466eeda1d6cd77d962c38e", "compiled_tokens": 323, "compiled_prompt_sha256": "aa541bf54e4da10cb7233c63ec2579c3ba492ff5efbefd219bec4731f7f5d42b", "compiled_ids_sha256": "6d3e88b7607381a1857be3378d363a3a49ad70080bd24153eee4a6f79fe0c29c", "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": "2026-09-28T16:44:49Z"} {"mode": "kev-choice", "prompt_sha256": "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "n_options": 6, "kev_response": {"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": 170.2576540010341, "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": 0.16639697299979161, "probabilities_f64_minus_runtime_max_abs": 2.368303819721973e-08}}}, "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": 0.17028217400002177, "prompt_tokens": 342, "eval_tokens": 1, "server_latency_ms": 170.2576540010341, "prompt_eval_ms": null, "total_ms": 170.28217400002177, "done_reason": null, "id": "c-ibn_sina-01-dessert-02", "uid": "c056", "task": "kc", "label": "ibn_sina", "text_sha256": "662296bb66a18a37300fd190cf48c9853e307386b2ca49c84cb14a90465ba367", "request_sha256": "ee3137cb7f1a66a6f4433d42bfd7c7a56e7450e35fcb2e2ab1ca53452e4d5595", "compiled_tokens": 342, "compiled_prompt_sha256": "725690d1f22c93d0340b285452fdb898cbeffbf60adbcee2a6cf9213e82abbc0", "compiled_ids_sha256": "94b95ec2aa38d8983aa6f7c9626041fbc139ec74d98175507b3c7f676e48c4ab", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "n_options": 6, "kev_response": {"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": 170.34300299928873, "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": 0.16652527199767064, "probabilities_f64_minus_runtime_max_abs": 6.906833294184622e-08}}}, "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": 0.1703674329983187, "prompt_tokens": 339, "eval_tokens": 1, "server_latency_ms": 170.34300299928873, "prompt_eval_ms": null, "total_ms": 170.3674329983187, "done_reason": null, "id": "c-ibn_sina-01-dessert-02", "uid": "c057", "task": "kc", "label": "ibn_sina", "text_sha256": "d8e0cb07e0eaf48824f57ebe5eb906dcef9a8f1e4b6f099dff399409791584b6", "request_sha256": "fef9b4a6bb369848c472d86f9cddd6d23a488ef8c4f0ddf47669b6b7d3933f4e", "compiled_tokens": 339, "compiled_prompt_sha256": "319fd600c27e869175120d52429454dc20b5548f94e0e370905b96ceb450ebfc", "compiled_ids_sha256": "7cd4b48291d7d001e70cad1045450e5ebc69a94b89e94cfc8233e6b63d6d448e", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "n_options": 6, "kev_response": {"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": 170.16907400102355, "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": 0.16623171500032186, "probabilities_f64_minus_runtime_max_abs": 1.5585478463364666e-08}}}, "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": 0.1701935040000535, "prompt_tokens": 351, "eval_tokens": 1, "server_latency_ms": 170.16907400102355, "prompt_eval_ms": null, "total_ms": 170.1935040000535, "done_reason": null, "id": "c-ibn_sina-01-dessert-03", "uid": "c058", "task": "kc", "label": "ibn_sina", "text_sha256": "f4471ae63e6d6bccbd5c261f4532ca1a299e3f457f483d8761a8918386e82bc1", "request_sha256": "a1a0666b3c7760ecf67320bc79e8c40eabd3038760e7b0d833c83d02d9857e7c", "compiled_tokens": 351, "compiled_prompt_sha256": "31e37b8f5f725ef736f2d61439d483950b0c0100444ccf92ee30ba189e9af0ef", "compiled_ids_sha256": "a7851da9192bd84a58d364630f143cf2398c247083217825c24b2c65fd8f0f9e", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "n_options": 6, "kev_response": {"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": 177.58522800068022, "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": 0.17326336100086337, "probabilities_f64_minus_runtime_max_abs": 2.1762632251842717e-08}}}, "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": 0.1776112679981452, "prompt_tokens": 393, "eval_tokens": 1, "server_latency_ms": 177.58522800068022, "prompt_eval_ms": null, "total_ms": 177.6112679981452, "done_reason": null, "id": "c-ibn_sina-01-dessert-03", "uid": "c059", "task": "kc", "label": "ibn_sina", "text_sha256": "bd105be757c9e850ab01e91c392d5b3ffcc1a669e30151f6297ad7d4a9f43e50", "request_sha256": "5d38405aad00196bf315b3f02a4541a96ee3b0b544e6d7b01ee62e4201af80de", "compiled_tokens": 393, "compiled_prompt_sha256": "ba26f55ee8938b9cb0c3d4aabab1559ed19fcddd55502d02f38ac6dfdc9310ac", "compiled_ids_sha256": "8f0bad5ac64183cef3d192f03fe419ee35d679774c7b48af85a668fb1670c955", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "n_options": 6, "kev_response": {"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": 165.74593800032744, "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": 0.16189422800016473, "probabilities_f64_minus_runtime_max_abs": 5.6967625061687954e-09}}}, "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": 0.1657727679994423, "prompt_tokens": 317, "eval_tokens": 1, "server_latency_ms": 165.74593800032744, "prompt_eval_ms": null, "total_ms": 165.7727679994423, "done_reason": null, "id": "c-ibn_sina-01-main-02", "uid": "c060", "task": "kc", "label": "ibn_sina", "text_sha256": "b55dbe714cc75405acc2e9cc1648dc3ee39c13f22e834fb80f8422ba653843e2", "request_sha256": "1a1a9d479c9f9382fc6b8f35f6aa832428b8ff9246586de0a5a16d82b97ed0d3", "compiled_tokens": 317, "compiled_prompt_sha256": "ca53ab37565565807844480c303f0ac73c53fed29285230b4137aee8eb216a3b", "compiled_ids_sha256": "46e60b1b41b20e85a4c648d7194e57c8c2da9d2e1ec5c2d9640a9a1b77e6ad6e", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "n_options": 6, "kev_response": {"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": 170.87137000271468, "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": 0.1669010799996613, "probabilities_f64_minus_runtime_max_abs": 1.674283045538516e-08}}}, "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": 0.17089730900261202, "prompt_tokens": 352, "eval_tokens": 1, "server_latency_ms": 170.87137000271468, "prompt_eval_ms": null, "total_ms": 170.89730900261202, "done_reason": null, "id": "c-ibn_sina-01-main-02", "uid": "c061", "task": "kc", "label": "ibn_sina", "text_sha256": "dd2f74b23716b02aa47bf6643e04dff6473ac0c4707db2e339a44e5a0a841efc", "request_sha256": "0eaa53cd7e68184de0b50355e7ae62c609c53927e1fff8511bf6410b513fd76c", "compiled_tokens": 352, "compiled_prompt_sha256": "45e5e8ec908f7291f32e80dc6bdc7bd7e15ac15096efb3d82a2feb0b9cc1ed9e", "compiled_ids_sha256": "8cb23d0f97d0241542c05f0a7bceae8684c28142b6f39ba5f8ef5db574533fac", "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": "2026-09-28T16:44:50Z"} {"mode": "kev-choice", "prompt_sha256": "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "n_options": 6, "kev_response": {"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": 170.72365100102616, "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": 0.16665351200208534, "probabilities_f64_minus_runtime_max_abs": 2.2214757339211744e-08}}}, "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": 0.17074872100056382, "prompt_tokens": 365, "eval_tokens": 1, "server_latency_ms": 170.72365100102616, "prompt_eval_ms": null, "total_ms": 170.74872100056382, "done_reason": null, "id": "c-ibn_sina-01-main-02", "uid": "c062", "task": "kc", "label": "ibn_sina", "text_sha256": "7017c582c8c06a8055ff39050ceb8b9c6ecdfd1950ea6fb82eaa12427352e499", "request_sha256": "c461c8ab53ff01b5a96bd7ece4257ee0a3381a919c1f261f77bf828da81d00de", "compiled_tokens": 365, "compiled_prompt_sha256": "2c19043fdf98431a6068bfeda9e535e1c58513f3ad737c128fbdbff9a34cc6d6", "compiled_ids_sha256": "3d9a463164b741f3095477f08090e3b5559a38fc08b00bb779a1cc9e5a019438", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "n_options": 6, "kev_response": {"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": 176.57391499960795, "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": 0.1721988399986003, "probabilities_f64_minus_runtime_max_abs": 2.5138659487389248e-08}}}, "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": 0.1766016449982999, "prompt_tokens": 390, "eval_tokens": 1, "server_latency_ms": 176.57391499960795, "prompt_eval_ms": null, "total_ms": 176.6016449982999, "done_reason": null, "id": "c-ibn_sina-01-main-03", "uid": "c063", "task": "kc", "label": "ibn_sina", "text_sha256": "0b11aa8d00f225a2e23695d1879b0b3cfecbb79f138d13ac59b948622fe38cae", "request_sha256": "9bf1321b83d4fa4a12fee84c88e428fa1366c4478d4eb1e3b333869cc146f6bf", "compiled_tokens": 390, "compiled_prompt_sha256": "7b9355b2e2fdcf6d623298737a222d53b12dc9d6c71de4bf927676f9c9b5c339", "compiled_ids_sha256": "45337ce0418c787e52b602dcd036092f44cf3be9bf63859bfa2bb7606fe58827", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "n_options": 6, "kev_response": {"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": 171.5762240019103, "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": 0.1673899059969699, "probabilities_f64_minus_runtime_max_abs": 2.6859842128246214e-08}}}, "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": 0.17160433400204056, "prompt_tokens": 371, "eval_tokens": 1, "server_latency_ms": 171.5762240019103, "prompt_eval_ms": null, "total_ms": 171.60433400204056, "done_reason": null, "id": "c-ibn_sina-01-main-03", "uid": "c064", "task": "kc", "label": "ibn_sina", "text_sha256": "a7d0bd3209b1bd2142727de630b760b15f5f6312081717f4e82e1ee2a0212263", "request_sha256": "eff8abe6388ec33a313b640330e601559224476208f89d3386053c3801501032", "compiled_tokens": 371, "compiled_prompt_sha256": "0dc80c8adce05b741ce8a4c3f17b9cc1f78bc491ba12ae7dfad74cd0ca49ecbb", "compiled_ids_sha256": "5948b85d5511809552d45e91339724803cfce0ca4a60973ed39e61cc07e354a8", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "n_options": 6, "kev_response": {"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": 178.07723400255782, "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": 0.17356319899772643, "probabilities_f64_minus_runtime_max_abs": 3.880197640970806e-08}}}, "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": 0.17810598299911362, "prompt_tokens": 395, "eval_tokens": 1, "server_latency_ms": 178.07723400255782, "prompt_eval_ms": null, "total_ms": 178.10598299911362, "done_reason": null, "id": "c-ibn_sina-01-main-03", "uid": "c065", "task": "kc", "label": "ibn_sina", "text_sha256": "3b833e5a5504cbd80f178ad75b526e1f554adfbb0089f158306ea21143c7ae2f", "request_sha256": "c5ad5e3e45e478abdf23a6ab5e1d56e3cd7dd44483f5f3db007ff88291281086", "compiled_tokens": 395, "compiled_prompt_sha256": "667abdaec59327957c3b1ce1c6570cf662d99312a7e39336fd34c6a5b9174833", "compiled_ids_sha256": "5cf614b764b9c0e5e91de93bb9bc44413b47819378f9850fb3926ac12e4ae96b", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "n_options": 6, "kev_response": {"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": 165.58172999793896, "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": 0.1620380469976226, "probabilities_f64_minus_runtime_max_abs": 6.110346151277923e-08}}}, "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": 0.16560614999980316, "prompt_tokens": 293, "eval_tokens": 1, "server_latency_ms": 165.58172999793896, "prompt_eval_ms": null, "total_ms": 165.60614999980316, "done_reason": null, "id": "c-ibn_sina-01-starter-01", "uid": "c066", "task": "kc", "label": "ibn_sina", "text_sha256": "23e51d7b6856050d91d3999166ecb603fc4961be70cc59ae3729087202ef3384", "request_sha256": "43311265711e35aea9915fd2616e6caa06d83a03a87905bc4b8ee72a6624f423", "compiled_tokens": 293, "compiled_prompt_sha256": "0cc652ed158d4456d057ae4ea6f240eebcd0b64c04d9666f8ff066084455a03d", "compiled_ids_sha256": "d87679564b3455f5e970ac7dffd32d0e50699c17f522bec77e89352be1475266", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "n_options": 6, "kev_response": {"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": 165.02222399867605, "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": 0.161512421000225, "probabilities_f64_minus_runtime_max_abs": 8.058490597040446e-09}}}, "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": 0.16504621399872121, "prompt_tokens": 311, "eval_tokens": 1, "server_latency_ms": 165.02222399867605, "prompt_eval_ms": null, "total_ms": 165.0462139987212, "done_reason": null, "id": "c-ibn_sina-01-starter-02", "uid": "c067", "task": "kc", "label": "ibn_sina", "text_sha256": "29e6791f3ab096493a37cc56ce2519a4c5cb87aafeed65dadcdbcbc34defc75c", "request_sha256": "131fc24eee14e0fad8c3f1bb63f44e6e9b71d0d8e5ad9dfd0ceb8f983efc446b", "compiled_tokens": 311, "compiled_prompt_sha256": "ecd96e4320ca6e089fea50dd88dd138efc7728b40394e645c167cf30bfe00de9", "compiled_ids_sha256": "ea289ecc97718490e97f10028ee7f3325d7a73f38345e7ea80efb30da1bcf2a9", "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": "2026-09-28T16:44:51Z"} {"mode": "kev-choice", "prompt_sha256": "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "n_options": 6, "kev_response": {"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": 171.25075600051787, "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": 0.16707995899923844, "probabilities_f64_minus_runtime_max_abs": 2.0168742453918753e-08}}}, "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": 0.17127617599908262, "prompt_tokens": 371, "eval_tokens": 1, "server_latency_ms": 171.25075600051787, "prompt_eval_ms": null, "total_ms": 171.27617599908262, "done_reason": null, "id": "c-ibn_sina-01-starter-03", "uid": "c068", "task": "kc", "label": "ibn_sina", "text_sha256": "92a317e3024d7f5dbd04f4abe6e6b2dbc9a2b7b76aef5d19ef117adbe1abff1f", "request_sha256": "34f4066062175b70b87d8fd7b2f00337e29a272948be077081e7cd483f164ddf", "compiled_tokens": 371, "compiled_prompt_sha256": "0ad9f34f9bb88bf2b729ba7edfdf37192f81ae6be3b2bb68a023efff32a6297e", "compiled_ids_sha256": "1fa8144179d5b9344965adbb89b17f9bf2ac697e24f6cd4b3cb376295fba968c", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "n_options": 6, "kev_response": {"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": 170.76977000033366, "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": 0.1666919909985154, "probabilities_f64_minus_runtime_max_abs": 4.906169048801701e-08}}}, "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": 0.17079722000198672, "prompt_tokens": 355, "eval_tokens": 1, "server_latency_ms": 170.76977000033366, "prompt_eval_ms": null, "total_ms": 170.79722000198672, "done_reason": null, "id": "c-ibn_sina-01-starter-03", "uid": "c069", "task": "kc", "label": "ibn_sina", "text_sha256": "2b38a3ae73d841c463a00f9961952498fec58ca7e4a8afc6ad0d68d3f2da0891", "request_sha256": "278ce52f1fb5806bd53426fc0d36fb0841720bd0af69b13c82171a1959ec2442", "compiled_tokens": 355, "compiled_prompt_sha256": "02824b55841c1a917666db92f2dab2f28df3699581fc36e4d28cc000f8f4a105", "compiled_ids_sha256": "052585d008f125e9a9c56574e6532af6ea4dffeaa060a8ab415742f7363867e1", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "n_options": 6, "kev_response": {"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": 170.8042789978208, "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": 0.16668836099779583, "probabilities_f64_minus_runtime_max_abs": 4.042546508653544e-08}}}, "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": 0.17083435899985489, "prompt_tokens": 371, "eval_tokens": 1, "server_latency_ms": 170.8042789978208, "prompt_eval_ms": null, "total_ms": 170.83435899985489, "done_reason": null, "id": "c-ibn_sina-01-starter-03", "uid": "c070", "task": "kc", "label": "ibn_sina", "text_sha256": "e7fbfb77c803a1b49ac58c6a9838ff600043b6d05103c23e581b0222b199442d", "request_sha256": "1ccbf400ac8a33b8f58b4d0235d5d5224724fdf4a03d6c8009022a3d8edce728", "compiled_tokens": 371, "compiled_prompt_sha256": "646ca9742583b77ea42ed581986481a7143ca23eac7d883e38862c3cb4e2dc56", "compiled_ids_sha256": "823c9e43923cf82515b47c19a4962b220c53c5e9d09cfb9b90c5cd78de0ad4e1", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "n_options": 6, "kev_response": {"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": 170.76374999669497, "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": 0.16688800000338233, "probabilities_f64_minus_runtime_max_abs": 6.928014162355112e-08}}}, "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": 0.1707945390007808, "prompt_tokens": 346, "eval_tokens": 1, "server_latency_ms": 170.76374999669497, "prompt_eval_ms": null, "total_ms": 170.7945390007808, "done_reason": null, "id": "c-ibn_sina-01-starter-03", "uid": "c071", "task": "kc", "label": "ibn_sina", "text_sha256": "75fffc308492668b7b9b1c55a6949b5364b630213ba04b50c512ebe89e45e583", "request_sha256": "ead36546ab9734fefac1b80c3330c225213caa33b146a86d9dab333bbe6d478c", "compiled_tokens": 346, "compiled_prompt_sha256": "672de484d211354cd558d5df735a439daa860248a900b0c7bbcb4c54d1e2256c", "compiled_ids_sha256": "1d298de3fe49f6c758af9b8bfaf818d27668637424e77a80f4a96f9bfce9406e", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "n_options": 6, "kev_response": {"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": 178.8156690017786, "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": 0.17438647199742263, "probabilities_f64_minus_runtime_max_abs": 9.142849921595086e-09}}}, "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": 0.1788425679987995, "prompt_tokens": 404, "eval_tokens": 1, "server_latency_ms": 178.8156690017786, "prompt_eval_ms": null, "total_ms": 178.8425679987995, "done_reason": null, "id": "c-sagan-01-dessert-01", "uid": "c072", "task": "kc", "label": "sagan", "text_sha256": "17954c8ac9ae5c7fedfad2b26338572afca1063b99a1e52048925e9d52e43d38", "request_sha256": "66aa66c272b747d081a054700b83be9e3bd9d52d659c476b286362026ac9636f", "compiled_tokens": 404, "compiled_prompt_sha256": "63461712757b659f4c1e81c96db955b9576be9a529dd979a61ae859b5e4fe20e", "compiled_ids_sha256": "82d8ecc5cf442b09faecb1a0ed1d65f9bc9153c91f2b0ea1a85d3471da004e30", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "n_options": 6, "kev_response": {"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": 165.5021709993889, "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": 0.16195306800000253, "probabilities_f64_minus_runtime_max_abs": 1.7207600344626428e-08}}}, "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": 0.16552666099960334, "prompt_tokens": 309, "eval_tokens": 1, "server_latency_ms": 165.5021709993889, "prompt_eval_ms": null, "total_ms": 165.52666099960334, "done_reason": null, "id": "c-sagan-01-dessert-01", "uid": "c073", "task": "kc", "label": "sagan", "text_sha256": "af4421e88e6f7cc45a04171042c0ae408960b8d2dd13e1e0b67b4579a5dafe83", "request_sha256": "985b8f250ac5989de5ec078b274393d5d7d4a02955eaad8be4423d5426ec6549", "compiled_tokens": 309, "compiled_prompt_sha256": "929aa8fbc3eee78c7018d5c8c61cbd18f10472b9ec10356a479f5560f36dc969", "compiled_ids_sha256": "9bfb6bfa3765f21c286543e6595240c7937c405557de6095941783fa77c32cf4", "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": "2026-09-28T16:44:52Z"} {"mode": "kev-choice", "prompt_sha256": "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "n_options": 6, "kev_response": {"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": 164.70523700263584, "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": 0.1613784119981574, "probabilities_f64_minus_runtime_max_abs": 3.558162542383059e-08}}}, "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": 0.16472792599961394, "prompt_tokens": 288, "eval_tokens": 1, "server_latency_ms": 164.70523700263584, "prompt_eval_ms": null, "total_ms": 164.72792599961394, "done_reason": null, "id": "c-sagan-01-dessert-01", "uid": "c074", "task": "kc", "label": "sagan", "text_sha256": "201a0135be6be1db753dcac7891bcbd6ebee66569e3359c08b3ee0a4105fd3fd", "request_sha256": "5ded00af7070527bfe7a3d8e16ee1bf2e306fa18d917c9c2a27b186fa4e3535d", "compiled_tokens": 288, "compiled_prompt_sha256": "c9162970d7a87671df926d763cd3c86cd736e6a98d9885f5541c96e5393b7873", "compiled_ids_sha256": "2e18cf1277f24da4cbb35e3c4d9ff0025b081d2a66b6e030656f384eefb90cf3", "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": "2026-09-28T16:44:53Z"} {"mode": "kev-choice", "prompt_sha256": "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "n_options": 6, "kev_response": {"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": 170.6086910016893, "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": 0.1669413000017812, "probabilities_f64_minus_runtime_max_abs": 5.285189708104099e-08}}}, "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": 0.17063462100122706, "prompt_tokens": 327, "eval_tokens": 1, "server_latency_ms": 170.6086910016893, "prompt_eval_ms": null, "total_ms": 170.63462100122706, "done_reason": null, "id": "c-sagan-01-dessert-03", "uid": "c075", "task": "kc", "label": "sagan", "text_sha256": "b17461d357df5f854916b406f705dfee2535c04fe6e0ec22055fb868d847f9a2", "request_sha256": "d7c43234b8433372fdf1337f8048ce306f6e3504c3cc8a7f19f1f7827196b61d", "compiled_tokens": 327, "compiled_prompt_sha256": "efc2df06e0745877c07b202c66d4cf238051ae0a1ae157bb1d319d831d699748", "compiled_ids_sha256": "6a8b4cc8d5bca66563e796c67461a643efc58f4871ab7302ea9ea36cd395825a", "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": "2026-09-28T16:44:53Z"} {"mode": "kev-choice", "prompt_sha256": "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "n_options": 6, "kev_response": {"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": 177.81580600058078, "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": 0.17306019199895673, "probabilities_f64_minus_runtime_max_abs": 2.5218937049764634e-08}}}, "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": 0.17784272599965334, "prompt_tokens": 409, "eval_tokens": 1, "server_latency_ms": 177.81580600058078, "prompt_eval_ms": null, "total_ms": 177.84272599965334, "done_reason": null, "id": "c-sagan-01-dessert-03", "uid": "c076", "task": "kc", "label": "sagan", "text_sha256": "0a559d120c34e49e361afce6d91c183b88fe89ed3166d332e2a0801f986305ea", "request_sha256": "16b9030c5ff61c8c38b3d8bfb022c702ef1463fba62f7e7ee16a0d11f701bcac", "compiled_tokens": 409, "compiled_prompt_sha256": "d1ab3fa14e7acc3c64152e9b9fefb12997ba7cb18617a7d4ba7df9edddd170aa", "compiled_ids_sha256": "b4fb76f1da222baeff1f762980a440de6c084e5a94b0a0af875b6114f77bd052", "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": "2026-09-28T16:44:53Z"} {"mode": "kev-choice", "prompt_sha256": "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "n_options": 6, "kev_response": {"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": 170.51826199895004, "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": 0.16685469999720226, "probabilities_f64_minus_runtime_max_abs": 9.911832399289722e-08}}}, "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": 0.1705425820000528, "prompt_tokens": 331, "eval_tokens": 1, "server_latency_ms": 170.51826199895004, "prompt_eval_ms": null, "total_ms": 170.5425820000528, "done_reason": null, "id": "c-sagan-01-dessert-03", "uid": "c077", "task": "kc", "label": "sagan", "text_sha256": "0a4333ec1c14f2d7026dd16373b6b7f5850ff30627e3985ba2719899cee3426f", "request_sha256": "6986af98fe6017d32e0ca28fc0acdaa56fd3a5c9618bb0433d6664286443c39a", "compiled_tokens": 331, "compiled_prompt_sha256": "5323ed669bcbd837510d737e7f7ff27ae1e5464e1a1530c9bd5f04eaf16e3def", "compiled_ids_sha256": "9fa1f31f937a41e05c35b71c72dada3a715a34d8f8cc835e0cc5cabd5ccf3784", "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": "2026-09-28T16:44:53Z"} {"mode": "kev-choice", "prompt_sha256": "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "n_options": 6, "kev_response": {"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": 173.6767670008703, "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": 0.16970858800050337, "probabilities_f64_minus_runtime_max_abs": 5.9411404129683376e-08}}}, "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": 0.1737025669972354, "prompt_tokens": 360, "eval_tokens": 1, "server_latency_ms": 173.6767670008703, "prompt_eval_ms": null, "total_ms": 173.7025669972354, "done_reason": null, "id": "c-sagan-01-dessert-03", "uid": "c078", "task": "kc", "label": "sagan", "text_sha256": "4adaea24337f52c5fd58eb48284a3be16efee7fd902865cb45b5695012a24e94", "request_sha256": "2b2181d544052beef5cbba60e7b4b98d112d0681cb2435766207b405755d7541", "compiled_tokens": 360, "compiled_prompt_sha256": "9185a3cc0cf1abb59a3142933ef89d11e19cf783575408d4218a74be6254ce2d", "compiled_ids_sha256": "36b52d662b922ffae58003f7a5d011a210ce57c6309ab056b16d4638019782b0", "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": "2026-09-28T16:44:53Z"} {"mode": "kev-choice", "prompt_sha256": "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "n_options": 6, "kev_response": {"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": 187.0651649987849, "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": 0.18266550800035475, "probabilities_f64_minus_runtime_max_abs": 7.239646215850826e-09}}}, "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": 0.1870926049996342, "prompt_tokens": 397, "eval_tokens": 1, "server_latency_ms": 187.0651649987849, "prompt_eval_ms": null, "total_ms": 187.0926049996342, "done_reason": null, "id": "c-sagan-01-dessert-03", "uid": "c079", "task": "kc", "label": "sagan", "text_sha256": "8c630a6e410940f6d22264df1640a7a2f0f4fe19eccbd45cbb1a2fee09efa8c2", "request_sha256": "fdf1072fd4e4239b84c8e15f11a2561f2bee8da5077fc190be78e7927dc93520", "compiled_tokens": 397, "compiled_prompt_sha256": "4c9fb986381beb2b96309a2afbd1b2af9fe10613bf4378bdb77d7479f8c41680", "compiled_ids_sha256": "86437e7858dab9217d982273312aa4f196d2a66f6de0c27248ed494a7c977df0", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "n_options": 6, "kev_response": {"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": 178.92393699730746, "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": 0.17519287599861855, "probabilities_f64_minus_runtime_max_abs": 2.050726732871766e-08}}}, "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": 0.17894976699972176, "prompt_tokens": 333, "eval_tokens": 1, "server_latency_ms": 178.92393699730746, "prompt_eval_ms": null, "total_ms": 178.94976699972176, "done_reason": null, "id": "c-sagan-01-main-01", "uid": "c080", "task": "kc", "label": "sagan", "text_sha256": "70f507eb1365d078f2ec073a7217f16108c3aa295d0f6f410b17376071934378", "request_sha256": "140f7548b8252bf789cc148366539da6583381dee3acb3f73e59965718c98631", "compiled_tokens": 333, "compiled_prompt_sha256": "9039ec9ebd536e8ff815ec4d3af1f3487d4b1b3ac68f56b8c74d168136f644a8", "compiled_ids_sha256": "d5b1938a23c4d5daa57a1a68304eaa2e03f3dfd41e38df0bac9ead15b4c8f182", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "n_options": 6, "kev_response": {"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": 171.062998000707, "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": 0.16696536900053616, "probabilities_f64_minus_runtime_max_abs": 2.238822793820816e-08}}}, "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": 0.17108907799774897, "prompt_tokens": 366, "eval_tokens": 1, "server_latency_ms": 171.062998000707, "prompt_eval_ms": null, "total_ms": 171.08907799774897, "done_reason": null, "id": "c-sagan-01-main-02", "uid": "c081", "task": "kc", "label": "sagan", "text_sha256": "26c9cc6cd1f5dfbf4575a10bfddada00c88453f4871b30ab1fc5c9322fa24311", "request_sha256": "365b36330a3df72c922382a9becff934f9afb952987aa40279f69f09eb7898a3", "compiled_tokens": 366, "compiled_prompt_sha256": "e899284aeb4ab20e1a02f56556620f4fbd0c64a5bfca3086c98d666d6ed9d79b", "compiled_ids_sha256": "baf3a4db52592f705e8840880143e8aa8842446449b21f0d838487828fea0bfd", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "n_options": 6, "kev_response": {"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": 165.49778999979026, "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": 0.1618705190012406, "probabilities_f64_minus_runtime_max_abs": 1.0558943497329665e-08}}}, "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": 0.1655228600029659, "prompt_tokens": 316, "eval_tokens": 1, "server_latency_ms": 165.49778999979026, "prompt_eval_ms": null, "total_ms": 165.5228600029659, "done_reason": null, "id": "c-sagan-01-main-02", "uid": "c082", "task": "kc", "label": "sagan", "text_sha256": "090da8cbfb7483c7bf328f01ab049ab77c06781715061f8708fd5d7b7edf51bc", "request_sha256": "dafefa8a14e06a9ddc266dbf3d063c14fef94ec25b3599ce335c518cc3b141fc", "compiled_tokens": 316, "compiled_prompt_sha256": "6b3262c0b05359df640a75374ab1cb9ee093c35d018b01ad37614f989d8ad180", "compiled_ids_sha256": "9cf18e6fda3d59410ee91dbb91dadbf48ceba6733de0f541fd3d4d6902aa5f4c", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "n_options": 6, "kev_response": {"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": 177.08247199698235, "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": 0.17262071599907358, "probabilities_f64_minus_runtime_max_abs": 2.387496145228596e-08}}}, "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": 0.17710995200104662, "prompt_tokens": 399, "eval_tokens": 1, "server_latency_ms": 177.08247199698235, "prompt_eval_ms": null, "total_ms": 177.10995200104662, "done_reason": null, "id": "c-sagan-01-main-03", "uid": "c083", "task": "kc", "label": "sagan", "text_sha256": "4ff2c7c47e30b6204ebc63d53fa463e60b4fc427332819e30a966cd466e5cf29", "request_sha256": "8f499288921bc7087ba6d4545f39cd67a4d51e8c55324c97852bc84f7e6cdcad", "compiled_tokens": 399, "compiled_prompt_sha256": "24b38735ab28fcd79f74277c7c0d9977f3ca86b28f20b968104f2e28ecd092e0", "compiled_ids_sha256": "c413e920def965f2042bf7df2f36f935761ed11a9ea1a07f8d6b709f90857e2d", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "n_options": 6, "kev_response": {"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": 178.16727299941704, "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": 0.17369993700049235, "probabilities_f64_minus_runtime_max_abs": 1.3312424856914973e-08}}}, "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": 0.17819566300022416, "prompt_tokens": 400, "eval_tokens": 1, "server_latency_ms": 178.16727299941704, "prompt_eval_ms": null, "total_ms": 178.19566300022416, "done_reason": null, "id": "c-sagan-01-main-03", "uid": "c084", "task": "kc", "label": "sagan", "text_sha256": "75fd209542847cbf68001d0bfbdc0906a9af020a3cae869a8c521fa0d7733d54", "request_sha256": "6fff42902f8b1b0e2028d0c6ecb1f95991fbdaf15e78d7d0fbaeb93b8ad7e726", "compiled_tokens": 400, "compiled_prompt_sha256": "18fc51ab32b6e239ab8298165343e7ff14bc85e0d166e6dc0b5e9964cae0940b", "compiled_ids_sha256": "73adbcdaa744c056b6a9cf97351cb9d64087fb324180be15d21ff25d3141cbfd", "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": "2026-09-28T16:44:54Z"} {"mode": "kev-choice", "prompt_sha256": "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "n_options": 6, "kev_response": {"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": 165.35462200045004, "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": 0.16216544700000668, "probabilities_f64_minus_runtime_max_abs": 5.3289591628313815e-09}}}, "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": 0.16537874099958572, "prompt_tokens": 284, "eval_tokens": 1, "server_latency_ms": 165.35462200045004, "prompt_eval_ms": null, "total_ms": 165.37874099958572, "done_reason": null, "id": "c-sagan-01-starter-01", "uid": "c085", "task": "kc", "label": "sagan", "text_sha256": "75430e1e6218b85b7b5a24b15accfed19fcb961b965741f729b8451ae554e6c3", "request_sha256": "895034ce1c9e079d133fa61f105a7d7186e0ba55df1f35ec288f4512b472d655", "compiled_tokens": 284, "compiled_prompt_sha256": "65b093d4dc7322a71d795f378d96f989360cca3b9cc8cc2dafc040037e482792", "compiled_ids_sha256": "c711190daf3352b61206e8801960f56744ba55154d91e3a74652ef38d6ba1604", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "n_options": 6, "kev_response": {"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": 177.17595100111794, "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": 0.17272242500257562, "probabilities_f64_minus_runtime_max_abs": 5.5442923252435605e-08}}}, "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": 0.17720187099985196, "prompt_tokens": 404, "eval_tokens": 1, "server_latency_ms": 177.17595100111794, "prompt_eval_ms": null, "total_ms": 177.20187099985196, "done_reason": null, "id": "c-sagan-01-starter-02", "uid": "c086", "task": "kc", "label": "sagan", "text_sha256": "75df625f87a54ffae2b7b0cdadfa23d16456e82b7c6e3d061a0358a590f5043f", "request_sha256": "30c7304be2a27ed3b3947ee0a061c75bc066da90f34275c77e6af9df4ea31263", "compiled_tokens": 404, "compiled_prompt_sha256": "e2b81005504d2eebe6b4f6eae69c57ba10486928d37eddc53c160a4b2854c594", "compiled_ids_sha256": "6f3ebc0833ff0d4eca3500ecdd2bf006ccd0dd84cbb8ec6be2d67191859098b9", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "n_options": 6, "kev_response": {"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": 165.1544729975285, "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": 0.1618320279994805, "probabilities_f64_minus_runtime_max_abs": 4.848229762099976e-08}}}, "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": 0.16517968300104258, "prompt_tokens": 293, "eval_tokens": 1, "server_latency_ms": 165.1544729975285, "prompt_eval_ms": null, "total_ms": 165.17968300104258, "done_reason": null, "id": "c-sagan-01-starter-02", "uid": "c087", "task": "kc", "label": "sagan", "text_sha256": "f37ad2a441d669b186a45bde189d7d209f5077324c7b15113285fce54185d55b", "request_sha256": "dc947051091fa4aa147b9543cf9fbe8970fe11bf085718543ba9583ec117426a", "compiled_tokens": 293, "compiled_prompt_sha256": "260bb13ff919a69d62276bf87d96d93384b67dfa42212c818966dc0bd3ee540c", "compiled_ids_sha256": "56677b4e63c956b87e199aa8482bbf7c197f78ec3d9d566703cd84fe0d74f83f", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "n_options": 6, "kev_response": {"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": 165.6625299983716, "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": 0.16222939499857603, "probabilities_f64_minus_runtime_max_abs": 2.0838092640307337e-08}}}, "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": 0.16568593900228734, "prompt_tokens": 302, "eval_tokens": 1, "server_latency_ms": 165.6625299983716, "prompt_eval_ms": null, "total_ms": 165.68593900228734, "done_reason": null, "id": "c-sagan-01-starter-03", "uid": "c088", "task": "kc", "label": "sagan", "text_sha256": "36439125e5a22ecbd8e1ade7eb23aeea2e65ec5efd46bd993882d91dd00c6cf1", "request_sha256": "ad9c564a4063a18ea249262eb0f3a4cc43a639316ae92fe79784687d6575409f", "compiled_tokens": 302, "compiled_prompt_sha256": "3824b179b4488ce489b4ef53aaa0bc5aac8823a7a8d88bcb5a7f0cf96ad47e1b", "compiled_ids_sha256": "4010bc1586d12cb94c392181608c684fece802cc25b6f26e7c35ffddfc2a80eb", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "n_options": 6, "kev_response": {"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": 170.7573899984709, "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": 0.1667857600004936, "probabilities_f64_minus_runtime_max_abs": 5.942152603299888e-09}}}, "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": 0.17078457000025082, "prompt_tokens": 355, "eval_tokens": 1, "server_latency_ms": 170.7573899984709, "prompt_eval_ms": null, "total_ms": 170.78457000025082, "done_reason": null, "id": "c-sagan-01-starter-03", "uid": "c089", "task": "kc", "label": "sagan", "text_sha256": "b82156a1523f861c6acec464a68044a0347260585b05e3d579f9d1bb6d3e1883", "request_sha256": "d5a506422f5cb8f1f1acbdb898efea3980d97e1926e0008e68ee1e8e095eaf1f", "compiled_tokens": 355, "compiled_prompt_sha256": "ce02283a565951bd24fed4ecdd1be1ff580fd4e71e1ed42c0be57924ef7735f0", "compiled_ids_sha256": "fcc59376315f837186e1531a6bc1a77307e5878970fb6cc3e97bcfa34479b565", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "n_options": 6, "kev_response": {"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": 177.41723900326178, "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": 0.17291154399936204, "probabilities_f64_minus_runtime_max_abs": 7.264001333773251e-08}}}, "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": 0.17744471799960593, "prompt_tokens": 409, "eval_tokens": 1, "server_latency_ms": 177.41723900326178, "prompt_eval_ms": null, "total_ms": 177.44471799960593, "done_reason": null, "id": "c-socrates-01-dessert-01", "uid": "c090", "task": "kc", "label": "socrates", "text_sha256": "b63520cf2b49ef5a68364a7ce70d03329569ecdd3852bcdf74b6daa63c3968bf", "request_sha256": "b5054215dae25e7dd6ed0d26aa4f538ffbc9f90176e00fecd90f25354f4f77ee", "compiled_tokens": 409, "compiled_prompt_sha256": "e5874d1b8d5df9f9ce009834e203a80ed9bfc65e1e117e65b805719bfafff002", "compiled_ids_sha256": "2fbf7e93347627cacd15188db8ccc02538cff99269ff3adb9d0ea548870e9cb9", "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": "2026-09-28T16:44:55Z"} {"mode": "kev-choice", "prompt_sha256": "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "n_options": 6, "kev_response": {"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": 166.55119300048682, "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": 0.16320530800294364, "probabilities_f64_minus_runtime_max_abs": 3.631840594930935e-08}}}, "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": 0.16657671299981303, "prompt_tokens": 291, "eval_tokens": 1, "server_latency_ms": 166.55119300048682, "prompt_eval_ms": null, "total_ms": 166.57671299981303, "done_reason": null, "id": "c-socrates-01-dessert-01", "uid": "c091", "task": "kc", "label": "socrates", "text_sha256": "f537b42e13f651d15d8bcfb648d10c7339118d0af15ef233f9a4976b21b55185", "request_sha256": "42e356c1a5a738bdceafa36690a1ba81cb1f60c2454be3eeb7763dc8bc966512", "compiled_tokens": 291, "compiled_prompt_sha256": "38c5218cbd49d0ad53554b844ea9581c36f8d9f6dab540ad7633c7aefaaec7f9", "compiled_ids_sha256": "2b8caa07c5125e82419f0a570ae0c7c4b742c0c93eaa8a8cd7f945e8e8e0643e", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "n_options": 6, "kev_response": {"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": 171.3125950009271, "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": 0.16733783699964988, "probabilities_f64_minus_runtime_max_abs": 1.2126818618729551e-08}}}, "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": 0.1713399849977577, "prompt_tokens": 361, "eval_tokens": 1, "server_latency_ms": 171.3125950009271, "prompt_eval_ms": null, "total_ms": 171.3399849977577, "done_reason": null, "id": "c-socrates-01-dessert-01", "uid": "c092", "task": "kc", "label": "socrates", "text_sha256": "58f8bc29dfc873fb65ccddb7a73ce568f962dc7f1f25e88e12dd0b0798caf239", "request_sha256": "67dcb6aab5956e3345ea02d36dd1493202dd845c88aa2d1ddf5c08d569e9e261", "compiled_tokens": 361, "compiled_prompt_sha256": "9a53e19f51ada7e3ca7748e01853f6e0dfe0321fc114aca34ef2b3bce186324a", "compiled_ids_sha256": "08a973dfb69a66733142afb8c93b88554f14300d2455d7ab0777f84c253ecdf1", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "n_options": 6, "kev_response": {"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": 173.258781000186, "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": 0.16943293100121082, "probabilities_f64_minus_runtime_max_abs": 1.3005094834372954e-08}}}, "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": 0.17328410100162728, "prompt_tokens": 343, "eval_tokens": 1, "server_latency_ms": 173.258781000186, "prompt_eval_ms": null, "total_ms": 173.28410100162728, "done_reason": null, "id": "c-socrates-01-dessert-02", "uid": "c093", "task": "kc", "label": "socrates", "text_sha256": "6b8746afe3db64759c4fe8e62f4167572b8707c6e0af2a303917c790e0107cf7", "request_sha256": "14b0361f2a076d27818b3a17225e24173e6c36324e06c05946a2103c8059563e", "compiled_tokens": 343, "compiled_prompt_sha256": "98bf941e9d49195d9f221459067344330d3de649eeabf47b96667525ca3aa283", "compiled_ids_sha256": "0f0f1ae75907c024b2d459061eabb496925526ca0d95c76d438b65fd51aa1e6e", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "n_options": 6, "kev_response": {"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": 175.11235699930694, "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": 0.17105564799931017, "probabilities_f64_minus_runtime_max_abs": 1.649327863173511e-08}}}, "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": 0.17513843699998688, "prompt_tokens": 374, "eval_tokens": 1, "server_latency_ms": 175.11235699930694, "prompt_eval_ms": null, "total_ms": 175.13843699998688, "done_reason": null, "id": "c-socrates-01-dessert-02", "uid": "c094", "task": "kc", "label": "socrates", "text_sha256": "91e20cb4e96bf48056541da6446b16d67d78651938fceace8dbfebd07fb6d458", "request_sha256": "d60dfdc0fb4ca19c34bc1382c9dd3b8883127f3d35673599556ccf58a577cec6", "compiled_tokens": 374, "compiled_prompt_sha256": "e0d5ffa591269263ed024ca529109dfb26c248d9d18c4a8d6e57f2ba9168b233", "compiled_ids_sha256": "42ef38f2cbdef5189c3d759a7044b4c8a748d18138cc275d8d40d3f3af6771d4", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "n_options": 6, "kev_response": {"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": 165.86215700226603, "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": 0.16248456299945246, "probabilities_f64_minus_runtime_max_abs": 3.5839498369050204e-08}}}, "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": 0.16588617700108443, "prompt_tokens": 302, "eval_tokens": 1, "server_latency_ms": 165.86215700226603, "prompt_eval_ms": null, "total_ms": 165.88617700108443, "done_reason": null, "id": "c-socrates-01-dessert-02", "uid": "c095", "task": "kc", "label": "socrates", "text_sha256": "9ffa8e8f85216a9cb308e9ac582dff3a3bef39e1d20109b23060822860780631", "request_sha256": "1926d0b3635e02750addb5db2c1806192a7739954e8245ae4a55fa4e482c4191", "compiled_tokens": 302, "compiled_prompt_sha256": "3b30247295b5c3dec8716911428c2b205adc94f0d63fb7499beab302bd9b5b24", "compiled_ids_sha256": "0da5ad6c304119d9f5f455fa27621e3de8086264734e81f84f183030922025e2", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "n_options": 6, "kev_response": {"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": 170.71554999711225, "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": 0.16710229800082743, "probabilities_f64_minus_runtime_max_abs": 8.736245615814653e-09}}}, "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": 0.17073967000033008, "prompt_tokens": 323, "eval_tokens": 1, "server_latency_ms": 170.71554999711225, "prompt_eval_ms": null, "total_ms": 170.73967000033008, "done_reason": null, "id": "c-socrates-01-main-01", "uid": "c096", "task": "kc", "label": "socrates", "text_sha256": "cdd1c49d12c260c7ebc9e05cb7f856333784a272c361416de516f780ca99b563", "request_sha256": "d89193cb5dc2ab9536b637ce578af9ac6c54d6ee49f7dab2728098cc245c3b79", "compiled_tokens": 323, "compiled_prompt_sha256": "1370e6b93e0997fcf22f6ff3e6473a4bccf01f168816a3e2da10413c14e6b44c", "compiled_ids_sha256": "a31f962774edf9fb5117efe523b6eb2f11ca98ce116e6d7c52e3734ea7a26aa9", "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": "2026-09-28T16:44:56Z"} {"mode": "kev-choice", "prompt_sha256": "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "n_options": 6, "kev_response": {"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": 178.09614399811835, "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": 0.17381112599832704, "probabilities_f64_minus_runtime_max_abs": 4.045006296582443e-08}}}, "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": 0.1781233039982908, "prompt_tokens": 395, "eval_tokens": 1, "server_latency_ms": 178.09614399811835, "prompt_eval_ms": null, "total_ms": 178.1233039982908, "done_reason": null, "id": "c-socrates-01-main-02", "uid": "c097", "task": "kc", "label": "socrates", "text_sha256": "cde59da3738cd9a8fe6df261efdaf3a4456b8bc09d08411d3000ce3a29c83b5b", "request_sha256": "b1bf9094b8e8971fa3e29b32555834ffa5d0e33970229230c1b2ce4615d93c8d", "compiled_tokens": 395, "compiled_prompt_sha256": "7a7cb526fa35845664c998ea567eaa951e2590b39bd2a7c881e769dec476a428", "compiled_ids_sha256": "fba0ef685096385bfc59e52dd6d585816ee5d2066e5bbea9c00a20c6d1da8968", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "n_options": 6, "kev_response": {"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": 165.31353200116428, "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": 0.16193707799902768, "probabilities_f64_minus_runtime_max_abs": 2.208841998196931e-08}}}, "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": 0.16533799200260546, "prompt_tokens": 297, "eval_tokens": 1, "server_latency_ms": 165.31353200116428, "prompt_eval_ms": null, "total_ms": 165.33799200260546, "done_reason": null, "id": "c-socrates-01-main-02", "uid": "c098", "task": "kc", "label": "socrates", "text_sha256": "7fcce46b50361f48ee0b98a7e17832e9023126cbeccd0326a9cbb0727f43f7ad", "request_sha256": "7b3f225227115a726e53cd581c75470b8f94abde5ff45d86102ab8830a619d53", "compiled_tokens": 297, "compiled_prompt_sha256": "ae667f183d359d40c186509ad74566d963a401022b071a698cabca10ad469305", "compiled_ids_sha256": "455a8afb9852c21fd5125d01820fb2e27419fa8e5daf8123220881908a4db405", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "n_options": 6, "kev_response": {"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": 176.0563589996309, "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": 0.1720901890003006, "probabilities_f64_minus_runtime_max_abs": 1.2529306592234946e-08}}}, "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": 0.17608129899963387, "prompt_tokens": 368, "eval_tokens": 1, "server_latency_ms": 176.0563589996309, "prompt_eval_ms": null, "total_ms": 176.08129899963387, "done_reason": null, "id": "c-socrates-01-main-02", "uid": "c099", "task": "kc", "label": "socrates", "text_sha256": "31b2c0bd67cd1c668ff7fbd043e45eec1df05d027787b72b24c896f2ef174f72", "request_sha256": "d88cd71424f272ecebd6dd7042e2a710895408bec112faac2f21b945c91ab755", "compiled_tokens": 368, "compiled_prompt_sha256": "241005cdbb0ac4de9754451ae5ce2a6d308deccb51feb8a49be378b2807af5d6", "compiled_ids_sha256": "7479577962a5718990d77f717d29e7cc80e2c6bc96a00d11d0d6eae9329107e2", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "n_options": 6, "kev_response": {"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": 169.57865900258184, "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": 0.1661253050006053, "probabilities_f64_minus_runtime_max_abs": 4.0938073153640175e-08}}}, "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": 0.16960227900199243, "prompt_tokens": 298, "eval_tokens": 1, "server_latency_ms": 169.57865900258184, "prompt_eval_ms": null, "total_ms": 169.60227900199243, "done_reason": null, "id": "c-socrates-01-main-02", "uid": "c100", "task": "kc", "label": "socrates", "text_sha256": "16c378fcf97035a1f0d659bd939e505f9e11f421fe386b8d29c5bed28d5fb98c", "request_sha256": "b3912f0c41f0fc5f8ba5bebb9305b5cb081aa57d59147bf2e054b9e3c92505e2", "compiled_tokens": 298, "compiled_prompt_sha256": "5a8c5648b9712c46350b863549c1ceed5801044f634662c32f1ac342b250aa4c", "compiled_ids_sha256": "b7bb323dfd3806fa1b596523b49ddfd4bc24913a34b61cf4015eef0afa9175a5", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "n_options": 6, "kev_response": {"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": 170.2824439998949, "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": 0.16656124200017075, "probabilities_f64_minus_runtime_max_abs": 4.994992919282737e-08}}}, "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": 0.17030691400213982, "prompt_tokens": 340, "eval_tokens": 1, "server_latency_ms": 170.2824439998949, "prompt_eval_ms": null, "total_ms": 170.30691400213982, "done_reason": null, "id": "c-socrates-01-main-03", "uid": "c101", "task": "kc", "label": "socrates", "text_sha256": "3f15b462d9b469045459958c57da06ec95d9f700f24d9477fff5a9b50ddd72aa", "request_sha256": "087772a504abc6edd03a515893d7186b659ed1ce62bcdf63c6380c519110815b", "compiled_tokens": 340, "compiled_prompt_sha256": "aa8a76f665cd82962eddc0c1cfe089528415b48ad3760ba5f5715122b5f5c9b1", "compiled_ids_sha256": "86a40936a666d5fd48b556179b1723278e9807945ccc3cb11a9396580f0dd5a4", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "n_options": 6, "kev_response": {"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": 171.1428269991302, "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": 0.16697083899998688, "probabilities_f64_minus_runtime_max_abs": 1.0811013195777264e-09}}}, "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": 0.1711678369974834, "prompt_tokens": 380, "eval_tokens": 1, "server_latency_ms": 171.1428269991302, "prompt_eval_ms": null, "total_ms": 171.1678369974834, "done_reason": null, "id": "c-socrates-01-main-03", "uid": "c102", "task": "kc", "label": "socrates", "text_sha256": "90e4b3b6a658a07636bbe827abfb381f93c6fb195ffd5ef085bf583f55af1d74", "request_sha256": "da39a310f76af967598a18569427bb9a79c40e6c5d35b8fd761b829c3d14d3fb", "compiled_tokens": 380, "compiled_prompt_sha256": "302cd7d856db5d188e298657e794f2abd5a5c6239133269be866a306600f9f17", "compiled_ids_sha256": "064b264cbea6255d0964374c8390710e5279938b074c351063f0f26916c2824f", "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": "2026-09-28T16:44:57Z"} {"mode": "kev-choice", "prompt_sha256": "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "n_options": 6, "kev_response": {"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": 165.2599420012848, "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": 0.16222133500195923, "probabilities_f64_minus_runtime_max_abs": 7.136863844348085e-08}}}, "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": 0.16528330199798802, "prompt_tokens": 267, "eval_tokens": 1, "server_latency_ms": 165.2599420012848, "prompt_eval_ms": null, "total_ms": 165.28330199798802, "done_reason": null, "id": "c-socrates-01-starter-01", "uid": "c103", "task": "kc", "label": "socrates", "text_sha256": "ce1f75337ffdd8f1d3134667d490376cc578923171abcb75d93c58afae704de3", "request_sha256": "ff6d1510cc4a9f868606a4497f9e4eb74877e054bdde5bac5ba03a41f299221c", "compiled_tokens": 267, "compiled_prompt_sha256": "7ba348fbd49cae74d977fc56e0a7c6c7e8609ef11e84d162768dc1a823298c56", "compiled_ids_sha256": "279db28cb15e54380a75c8b3d192304b55a16372b39c8bf6c63f78d6061d4052", "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": "2026-09-28T16:44:58Z"} {"mode": "kev-choice", "prompt_sha256": "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "n_options": 6, "kev_response": {"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": 166.13058499933686, "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": 0.1625669029999699, "probabilities_f64_minus_runtime_max_abs": 3.379026813910002e-08}}}, "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": 0.16615495500082034, "prompt_tokens": 315, "eval_tokens": 1, "server_latency_ms": 166.13058499933686, "prompt_eval_ms": null, "total_ms": 166.15495500082034, "done_reason": null, "id": "c-socrates-01-starter-02", "uid": "c104", "task": "kc", "label": "socrates", "text_sha256": "d2544f4722f7953ab187db9c08752f507f9e7bd525b074e990730d32763f93b0", "request_sha256": "caa5afacee5ba562467873a4832b48d03710f29e44cf8989baea421877ce095d", "compiled_tokens": 315, "compiled_prompt_sha256": "09a3150ed8b79aaef3b11c5a10df28e718cebf9189825e56bd6872be31be436d", "compiled_ids_sha256": "2fae6a509382acdcb1398799443b33dcf79f533516bd7b6f1ae007a1d10494dc", "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": "2026-09-28T16:44:58Z"} {"mode": "kev-choice", "prompt_sha256": "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "n_options": 6, "kev_response": {"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": 170.79339999691, "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": 0.16714915800184826, "probabilities_f64_minus_runtime_max_abs": 4.4798891596364854e-08}}}, "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": 0.17081770000004326, "prompt_tokens": 335, "eval_tokens": 1, "server_latency_ms": 170.79339999691, "prompt_eval_ms": null, "total_ms": 170.81770000004326, "done_reason": null, "id": "c-socrates-01-starter-03", "uid": "c105", "task": "kc", "label": "socrates", "text_sha256": "76bd6a400f7300d59d745ae87514d123122ad9184a8cbf280deb317001c2365e", "request_sha256": "b85349448634763b9f05d811ce1c6f89af68fc4dc67dea4417ef46bb2e66e79f", "compiled_tokens": 335, "compiled_prompt_sha256": "1b43313b82313432c0e61c8d24490e5ee29de32faa2f8898725bc668a978fa6f", "compiled_ids_sha256": "8635156608b5843ff3e2d0a3a4e35ad222d413b6cce00d96311035874e998319", "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": "2026-09-28T16:44:58Z"} {"mode": "kev-choice", "prompt_sha256": "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "n_options": 6, "kev_response": {"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": 170.8165790005296, "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": 0.16688568999961717, "probabilities_f64_minus_runtime_max_abs": 8.012177876715043e-08}}}, "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": 0.17084369900112506, "prompt_tokens": 355, "eval_tokens": 1, "server_latency_ms": 170.8165790005296, "prompt_eval_ms": null, "total_ms": 170.84369900112506, "done_reason": null, "id": "c-socrates-01-starter-03", "uid": "c106", "task": "kc", "label": "socrates", "text_sha256": "62ebf3000b7bc1337e703d293f168f2fef0e77979b30055c111974f6d491f990", "request_sha256": "4de53544ee23e1e93fd3d21ab2a5290ccc1d0e8d4141479fccab952a097e4269", "compiled_tokens": 355, "compiled_prompt_sha256": "e89581a53f340a2aeeb9384cd75a6f0aefe0ea16e3efc03c033b3b5f3f2847e8", "compiled_ids_sha256": "67d6cefb26c2e1a4ee433d8b03c2659df0da773a4de7d34f23c8e3cc19bf6703", "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": "2026-09-28T16:44:58Z"} {"mode": "kev-choice", "prompt_sha256": "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "n_options": 6, "kev_response": {"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": 170.92227899775025, "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": 0.1672822369982896, "probabilities_f64_minus_runtime_max_abs": 4.135007691807857e-08}}}, "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": 0.17094730800090474, "prompt_tokens": 328, "eval_tokens": 1, "server_latency_ms": 170.92227899775025, "prompt_eval_ms": null, "total_ms": 170.94730800090474, "done_reason": null, "id": "c-socrates-01-starter-03", "uid": "c107", "task": "kc", "label": "socrates", "text_sha256": "d9b9706fa1986a4b001053dac8268bae4e0be7b3860099d80287a3aab18de4e2", "request_sha256": "0f756818bdff4088317d34f8aedc6ab0883679e7961bf2faa0d0a75e163f2296", "compiled_tokens": 328, "compiled_prompt_sha256": "8501d29a2806beb81cce7eb55e95f965f1f81dc43c69104f947a1d2360bdb71d", "compiled_ids_sha256": "da1348b4fbd3b223210453e72685938201ef08da043c35159863b5490ea36eae", "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": "2026-09-28T16:44:58Z"}