{ "arm": "num_thread sweep (full scoring rules: 1 warmup + N scored, median)", "box_class": "cpu-laptop", "model_tag": "qwen3.5:9b", "nproc": 24, "runs_per_setting": 5, "rows": [ { "num_thread": 4, "decode_median": 7.409, "decode_min": 7.36, "decode_max": 7.413, "loadavg_median": 3.7 }, { "num_thread": 8, "decode_median": 10.484, "decode_min": 10.479, "decode_max": 10.526, "loadavg_median": 7.18 }, { "num_thread": 12, "decode_median": 12.165, "decode_min": 11.935, "decode_max": 12.169, "loadavg_median": 11.1 }, { "num_thread": 16, "decode_median": 12.819, "decode_min": 12.721, "decode_max": 12.887, "loadavg_median": 14.57 }, { "num_thread": 20, "decode_median": 13.048, "decode_min": 13.033, "decode_max": 13.314, "loadavg_median": 18.77 }, { "num_thread": 24, "decode_median": 11.984, "decode_min": 10.273, "decode_max": 12.722, "loadavg_median": 23.47 } ], "best_threads": 20, "detail": [ { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=4", "num_thread": 4, "started_utc": "2026-08-25T18:01:20Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "0.28 0.33 0.71 1/1408 2857095", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 259.74, "first_token_was_thinking": false, "wall_s": 35.0432, "eval_count": 256, "eval_duration_ns": 34781674000, "decode_tok_s": 7.36, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2106.965, "load_duration_ns": 818424, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 258.72, "first_token_was_thinking": false, "wall_s": 34.9999, "eval_count": 256, "eval_duration_ns": 34737910000, "decode_tok_s": 7.369, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2113.749, "load_duration_ns": 842392, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 256.05, "first_token_was_thinking": false, "wall_s": 34.8087, "eval_count": 256, "eval_duration_ns": 34550472000, "decode_tok_s": 7.409, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2125.192, "load_duration_ns": 744928, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 259.84, "first_token_was_thinking": false, "wall_s": 34.8042, "eval_count": 256, "eval_duration_ns": 34539823000, "decode_tok_s": 7.412, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2126.387, "load_duration_ns": 989804, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 259.33, "first_token_was_thinking": false, "wall_s": 34.793, "eval_count": 256, "eval_duration_ns": 34532108000, "decode_tok_s": 7.413, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2117.141, "load_duration_ns": 1473913, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 4, "ttft_ms_CACHED_PREFIX": 259.09, "first_token_was_thinking": false, "wall_s": 35.0199, "eval_count": 256, "eval_duration_ns": 34757462000, "decode_tok_s": 7.365, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 2120.585, "load_duration_ns": 779084, "streamed_chunks": 239, "response_chars": 1040, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:12:38.225037277Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.64, "peak_runner_threads": 31, "loadavg_during_runs": { "min": 2.51, "median": 3.7, "max": 4.0, "mean": 3.56, "samples": 341, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 7.36, "median": 7.409, "max": 7.413, "mean": 7.393, "n": 5, "spread_pct": 0.72 }, "wall_s": { "min": 34.793, "median": 34.8087, "max": 35.0432, "mean": 34.89, "n": 5, "spread_pct": 0.72 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:05:33Z" }, { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=8", "num_thread": 8, "started_utc": "2026-08-25T18:05:33Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "3.67 2.28 1.43 1/1412 2860253", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 165.77, "first_token_was_thinking": false, "wall_s": 24.4897, "eval_count": 256, "eval_duration_ns": 24320065000, "decode_tok_s": 10.526, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3323.781, "load_duration_ns": 673654, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 168.84, "first_token_was_thinking": false, "wall_s": 24.5593, "eval_count": 256, "eval_duration_ns": 24386958000, "decode_tok_s": 10.497, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3286.132, "load_duration_ns": 872814, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 169.25, "first_token_was_thinking": false, "wall_s": 24.5927, "eval_count": 256, "eval_duration_ns": 24419715000, "decode_tok_s": 10.483, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3300.675, "load_duration_ns": 1197529, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 169.17, "first_token_was_thinking": false, "wall_s": 24.5918, "eval_count": 256, "eval_duration_ns": 24418877000, "decode_tok_s": 10.484, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3300.939, "load_duration_ns": 987631, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 170.12, "first_token_was_thinking": false, "wall_s": 24.6032, "eval_count": 256, "eval_duration_ns": 24430893000, "decode_tok_s": 10.479, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3295.51, "load_duration_ns": 1201136, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 8, "ttft_ms_CACHED_PREFIX": 162.27, "first_token_was_thinking": false, "wall_s": 24.4715, "eval_count": 256, "eval_duration_ns": 24305293000, "decode_tok_s": 10.533, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3423.5, "load_duration_ns": 1170919, "streamed_chunks": 237, "response_chars": 1021, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:16:10.262702644Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.65, "peak_runner_threads": 35, "loadavg_during_runs": { "min": 5.59, "median": 7.18, "max": 7.89, "mean": 7.13, "samples": 240, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 10.479, "median": 10.484, "max": 10.526, "mean": 10.494, "n": 5, "spread_pct": 0.45 }, "wall_s": { "min": 24.4897, "median": 24.5918, "max": 24.6032, "mean": 24.567, "n": 5, "spread_pct": 0.46 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:08:13Z" }, { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=12", "num_thread": 12, "started_utc": "2026-08-25T18:08:13Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "7.89 4.74 2.51 1/1405 2861794", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 139.98, "first_token_was_thinking": false, "wall_s": 21.1804, "eval_count": 256, "eval_duration_ns": 21038391000, "decode_tok_s": 12.168, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3968.049, "load_duration_ns": 612753, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 140.76, "first_token_was_thinking": false, "wall_s": 21.1871, "eval_count": 256, "eval_duration_ns": 21044076000, "decode_tok_s": 12.165, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3942.123, "load_duration_ns": 1031788, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 140.24, "first_token_was_thinking": false, "wall_s": 21.1802, "eval_count": 256, "eval_duration_ns": 21037803000, "decode_tok_s": 12.169, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3959.856, "load_duration_ns": 554970, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 142.12, "first_token_was_thinking": false, "wall_s": 21.2206, "eval_count": 256, "eval_duration_ns": 21076858000, "decode_tok_s": 12.146, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3919.651, "load_duration_ns": 583200, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 140.78, "first_token_was_thinking": false, "wall_s": 21.5923, "eval_count": 256, "eval_duration_ns": 21448901000, "decode_tok_s": 11.935, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3913.139, "load_duration_ns": 585986, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 12, "ttft_ms_CACHED_PREFIX": 140.08, "first_token_was_thinking": false, "wall_s": 21.3102, "eval_count": 256, "eval_duration_ns": 21167963000, "decode_tok_s": 12.094, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3982.232, "load_duration_ns": 865084, "streamed_chunks": 238, "response_chars": 1017, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:18:45.702755212Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.65, "peak_runner_threads": 39, "loadavg_during_runs": { "min": 9.51, "median": 11.1, "max": 11.61, "mean": 10.91, "samples": 208, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 11.935, "median": 12.165, "max": 12.169, "mean": 12.117, "n": 5, "spread_pct": 1.92 }, "wall_s": { "min": 21.1802, "median": 21.1871, "max": 21.5923, "mean": 21.272, "n": 5, "spread_pct": 1.95 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:10:32Z" }, { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=16", "num_thread": 16, "started_utc": "2026-08-25T18:10:32Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "11.61 7.41 3.81 1/1410 2863688", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 131.66, "first_token_was_thinking": false, "wall_s": 19.9984, "eval_count": 256, "eval_duration_ns": 19864765000, "decode_tok_s": 12.887, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4265.798, "load_duration_ns": 1001844, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 132.36, "first_token_was_thinking": false, "wall_s": 20.1049, "eval_count": 256, "eval_duration_ns": 19970196000, "decode_tok_s": 12.819, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4210.906, "load_duration_ns": 876873, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 132.97, "first_token_was_thinking": false, "wall_s": 20.0209, "eval_count": 256, "eval_duration_ns": 19886673000, "decode_tok_s": 12.873, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4215.269, "load_duration_ns": 571359, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 133.85, "first_token_was_thinking": false, "wall_s": 20.1831, "eval_count": 256, "eval_duration_ns": 20046995000, "decode_tok_s": 12.77, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4134.204, "load_duration_ns": 672541, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 135.02, "first_token_was_thinking": false, "wall_s": 20.2608, "eval_count": 256, "eval_duration_ns": 20123693000, "decode_tok_s": 12.721, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4158.31, "load_duration_ns": 555078, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 16, "ttft_ms_CACHED_PREFIX": 130.59, "first_token_was_thinking": false, "wall_s": 19.693, "eval_count": 256, "eval_duration_ns": 19560327000, "decode_tok_s": 13.088, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4290.44, "load_duration_ns": 769788, "streamed_chunks": 238, "response_chars": 1048, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:21:02.432278064Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.65, "peak_runner_threads": 43, "loadavg_during_runs": { "min": 12.55, "median": 14.57, "max": 15.44, "mean": 14.43, "samples": 196, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 12.721, "median": 12.819, "max": 12.887, "mean": 12.814, "n": 5, "spread_pct": 1.29 }, "wall_s": { "min": 19.9984, "median": 20.1049, "max": 20.2608, "mean": 20.114, "n": 5, "spread_pct": 1.31 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:12:43Z" }, { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=20", "num_thread": 20, "started_utc": "2026-08-25T18:12:43Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "15.49 10.41 5.40 1/1406 2865062", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 117.2, "first_token_was_thinking": false, "wall_s": 19.3479, "eval_count": 256, "eval_duration_ns": 19228401000, "decode_tok_s": 13.314, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4830.961, "load_duration_ns": 519699, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 118.16, "first_token_was_thinking": false, "wall_s": 19.5703, "eval_count": 256, "eval_duration_ns": 19450330000, "decode_tok_s": 13.162, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4792.888, "load_duration_ns": 579924, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 123.01, "first_token_was_thinking": false, "wall_s": 19.7564, "eval_count": 256, "eval_duration_ns": 19630659000, "decode_tok_s": 13.041, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4679.127, "load_duration_ns": 2024097, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 118.77, "first_token_was_thinking": false, "wall_s": 19.7641, "eval_count": 256, "eval_duration_ns": 19642506000, "decode_tok_s": 13.033, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4741.387, "load_duration_ns": 528075, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 116.83, "first_token_was_thinking": false, "wall_s": 19.739, "eval_count": 256, "eval_duration_ns": 19620166000, "decode_tok_s": 13.048, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4761.103, "load_duration_ns": 470171, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 20, "ttft_ms_CACHED_PREFIX": 117.67, "first_token_was_thinking": false, "wall_s": 19.2989, "eval_count": 256, "eval_duration_ns": 19178303000, "decode_tok_s": 13.348, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4800.815, "load_duration_ns": 843800, "streamed_chunks": 239, "response_chars": 1057, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:23:12.265123531Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.65, "peak_runner_threads": 47, "loadavg_during_runs": { "min": 17.03, "median": 18.77, "max": 19.81, "mean": 18.71, "samples": 190, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 13.033, "median": 13.048, "max": 13.314, "mean": 13.12, "n": 5, "spread_pct": 2.15 }, "wall_s": { "min": 19.3479, "median": 19.739, "max": 19.7641, "mean": 19.636, "n": 5, "spread_pct": 2.11 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:14:50Z" }, { "model_id": "m1", "model_tag": "qwen3.5:9b", "box_class": "cpu-laptop", "arm_label": "sweep t=24", "num_thread": 24, "started_utc": "2026-08-25T18:14:50Z", "nproc": 24, "num_predict": 256, "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c", "loadavg_before_arm": "19.81 13.80 7.29 1/1404 2866412", "nvidia_before": "0, 15 MiB, 0 %", "ollama_version": "0.32.15", "runs": [ { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 120.18, "first_token_was_thinking": false, "wall_s": 20.2445, "eval_count": 256, "eval_duration_ns": 20122473000, "decode_tok_s": 12.722, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4692.784, "load_duration_ns": 1055178, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length", "run": 1 }, { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 112.64, "first_token_was_thinking": false, "wall_s": 20.3111, "eval_count": 256, "eval_duration_ns": 20196498000, "decode_tok_s": 12.675, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4873.71, "load_duration_ns": 451123, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length", "run": 2 }, { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 112.08, "first_token_was_thinking": false, "wall_s": 25.0367, "eval_count": 256, "eval_duration_ns": 24919436000, "decode_tok_s": 10.273, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 5058.45, "load_duration_ns": 401330, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length", "run": 3 }, { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 143.04, "first_token_was_thinking": false, "wall_s": 21.5053, "eval_count": 256, "eval_duration_ns": 21361766000, "decode_tok_s": 11.984, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 3960.235, "load_duration_ns": 1988629, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length", "run": 4 }, { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 115.6, "first_token_was_thinking": false, "wall_s": 22.3988, "eval_count": 256, "eval_duration_ns": 22282774000, "decode_tok_s": 11.489, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 4981.586, "load_duration_ns": 1308452, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length", "run": 5 } ], "errors": [], "quantization": "Q4_K_M", "param_size": "9.7B", "context_length": 262144, "think_mode": false, "think_note": "think:false accepted", "warmup_discarded": { "num_thread": 24, "ttft_ms_CACHED_PREFIX": 111.06, "first_token_was_thinking": false, "wall_s": 21.1606, "eval_count": 256, "eval_duration_ns": 21049926000, "decode_tok_s": 12.162, "prompt_eval_count": 537, "prefill_tok_s_CACHE_ARTIFACT": 5158.303, "load_duration_ns": 845866, "streamed_chunks": 239, "response_chars": 1041, "thinking_chars": 0, "done_reason": "length" }, "ps_after_load": [ { "name": "qwen3.5:9b", "model": "qwen3.5:9b", "size": 6071601067, "digest": "6488c96fa5faab64bb65cbd30d4289e20e6130ef535a93ef9a49f42eda893ea7", "details": { "parent_model": "", "format": "gguf", "family": "qwen35", "families": [ "qwen35" ], "parameter_size": "9.7B", "quantization_level": "Q4_K_M" }, "expires_at": "2026-08-25T18:25:22.258495293Z", "size_vram": 0, "context_length": 4096 } ], "size_vram": 0, "size_total": 6071601067, "cpu_only_receipt": true, "peak_runner_rss_gib": 6.64, "peak_runner_threads": 51, "loadavg_during_runs": { "min": 20.42, "median": 23.47, "max": 24.94, "mean": 23.19, "samples": 210, "nproc": 24 }, "contention_suspected": false, "nvidia_after": "0, 15 MiB, 0 %", "derived": { "decode_tok_s": { "min": 10.273, "median": 11.984, "max": 12.722, "mean": 11.829, "n": 5, "spread_pct": 20.44 }, "wall_s": { "min": 20.2445, "median": 21.5053, "max": 25.0367, "mean": 21.899, "n": 5, "spread_pct": 22.28 }, "eval_count": { "min": 256, "median": 256, "max": 256, "mean": 256.0, "n": 5, "spread_pct": 0.0 }, "thinking_chars_total": 0 }, "finished_utc": "2026-08-25T18:17:11Z" } ], "finished_utc": "2026-08-25T18:17:12Z" }