{
  "arm": "num_thread sweep (full scoring rules: 1 warmup + N scored, median)",
  "box_class": "cpu-mini",
  "model_tag": "nemotron-3.5-lightning:30b-a3b",
  "nproc": 16,
  "runs_per_setting": 5,
  "rows": [
    {
      "num_thread": 4,
      "decode_median": 14.436,
      "decode_min": 14.139,
      "decode_max": 15.424,
      "loadavg_median": 4.08
    },
    {
      "num_thread": 6,
      "decode_median": 13.604,
      "decode_min": 13.602,
      "decode_max": 13.631,
      "loadavg_median": 5.65
    },
    {
      "num_thread": 8,
      "decode_median": 12.524,
      "decode_min": 11.681,
      "decode_max": 13.377,
      "loadavg_median": 9.92
    },
    {
      "num_thread": 12,
      "decode_median": 7.28,
      "decode_min": 7.203,
      "decode_max": 7.353,
      "loadavg_median": 16.41
    },
    {
      "num_thread": 16,
      "decode_median": 1.273,
      "decode_min": 1.238,
      "decode_max": 1.432,
      "loadavg_median": 19.02
    }
  ],
  "best_threads": 4,
  "detail": [
    {
      "model_id": "m6",
      "model_tag": "nemotron-3.5-lightning:30b-a3b",
      "box_class": "cpu-mini",
      "arm_label": "sweep t=4",
      "num_thread": 4,
      "started_utc": "2026-08-25T19:14:25Z",
      "nproc": 16,
      "num_predict": 256,
      "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c",
      "loadavg_before_arm": "2.88 3.90 5.13 1/324 81119",
      "nvidia_before": "no nvidia-smi (no GPU on this box)",
      "ollama_version": "0.32.15",
      "runs": [
        {
          "num_thread": 4,
          "ttft_ms_CACHED_PREFIX": 184.03,
          "first_token_was_thinking": false,
          "wall_s": 18.2913,
          "eval_count": 256,
          "eval_duration_ns": 18105720000,
          "decode_tok_s": 14.139,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2954.882,
          "load_duration_ns": 873877,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 1
        },
        {
          "num_thread": 4,
          "ttft_ms_CACHED_PREFIX": 184.64,
          "first_token_was_thinking": false,
          "wall_s": 17.9194,
          "eval_count": 256,
          "eval_duration_ns": 17733295000,
          "decode_tok_s": 14.436,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2939.275,
          "load_duration_ns": 842649,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 2
        },
        {
          "num_thread": 4,
          "ttft_ms_CACHED_PREFIX": 170.31,
          "first_token_was_thinking": false,
          "wall_s": 17.9522,
          "eval_count": 256,
          "eval_duration_ns": 17780358000,
          "decode_tok_s": 14.398,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 3192.377,
          "load_duration_ns": 925337,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 3
        },
        {
          "num_thread": 4,
          "ttft_ms_CACHED_PREFIX": 170.14,
          "first_token_was_thinking": false,
          "wall_s": 17.4985,
          "eval_count": 256,
          "eval_duration_ns": 17326882000,
          "decode_tok_s": 14.775,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 3199.846,
          "load_duration_ns": 919407,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 4
        },
        {
          "num_thread": 4,
          "ttft_ms_CACHED_PREFIX": 172.63,
          "first_token_was_thinking": false,
          "wall_s": 16.7722,
          "eval_count": 256,
          "eval_duration_ns": 16598043000,
          "decode_tok_s": 15.424,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 3155.547,
          "load_duration_ns": 920800,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 5
        }
      ],
      "errors": [],
      "quantization": "Q4_K_M",
      "param_size": "32.9B",
      "context_length": 1048576,
      "think_mode": false,
      "think_note": "think:false accepted",
      "warmup_discarded": {
        "num_thread": 4,
        "ttft_ms_CACHED_PREFIX": 181.6,
        "first_token_was_thinking": false,
        "wall_s": 18.0594,
        "eval_count": 256,
        "eval_duration_ns": 17876007000,
        "decode_tok_s": 14.321,
        "prompt_eval_count": 532,
        "prefill_tok_s_CACHE_ARTIFACT": 2990.444,
        "load_duration_ns": 869375,
        "streamed_chunks": 245,
        "response_chars": 1029,
        "thinking_chars": 0,
        "done_reason": "length"
      },
      "ps_after_load": [
        {
          "name": "nemotron-3.5-lightning:30b-a3b",
          "model": "nemotron-3.5-lightning:30b-a3b",
          "size": 25609340188,
          "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2",
          "details": {
            "parent_model": "",
            "format": "gguf",
            "family": "nemotron_h_moe",
            "families": [
              "nemotron_h_moe"
            ],
            "parameter_size": "32.9B",
            "quantization_level": "Q4_K_M"
          },
          "expires_at": "2026-08-25T19:25:07.656651913Z",
          "size_vram": 0,
          "context_length": 4096
        }
      ],
      "size_vram": 0,
      "size_total": 25609340188,
      "cpu_only_receipt": true,
      "peak_runner_rss_gib": 24.17,
      "peak_runner_threads": 26,
      "loadavg_during_runs": {
        "min": 3.61,
        "median": 4.08,
        "max": 4.15,
        "mean": 3.98,
        "samples": 174,
        "nproc": 16
      },
      "contention_suspected": false,
      "nvidia_after": "no nvidia-smi (no GPU on this box)",
      "derived": {
        "decode_tok_s": {
          "min": 14.139,
          "median": 14.436,
          "max": 15.424,
          "mean": 14.634,
          "n": 5,
          "spread_pct": 8.9
        },
        "wall_s": {
          "min": 16.7722,
          "median": 17.9194,
          "max": 18.2913,
          "mean": 17.687,
          "n": 5,
          "spread_pct": 8.48
        },
        "eval_count": {
          "min": 256,
          "median": 256,
          "max": 256,
          "mean": 256.0,
          "n": 5,
          "spread_pct": 0.0
        },
        "thinking_chars_total": 0
      },
      "finished_utc": "2026-08-25T19:16:36Z"
    },
    {
      "model_id": "m6",
      "model_tag": "nemotron-3.5-lightning:30b-a3b",
      "box_class": "cpu-mini",
      "arm_label": "sweep t=6",
      "num_thread": 6,
      "started_utc": "2026-08-25T19:16:36Z",
      "nproc": 16,
      "num_predict": 256,
      "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c",
      "loadavg_before_arm": "4.06 4.01 5.00 1/324 86895",
      "nvidia_before": "no nvidia-smi (no GPU on this box)",
      "ollama_version": "0.32.15",
      "runs": [
        {
          "num_thread": 6,
          "ttft_ms_CACHED_PREFIX": 209.07,
          "first_token_was_thinking": false,
          "wall_s": 18.9909,
          "eval_count": 256,
          "eval_duration_ns": 18780175000,
          "decode_tok_s": 13.631,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2590.61,
          "load_duration_ns": 889437,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 1
        },
        {
          "num_thread": 6,
          "ttft_ms_CACHED_PREFIX": 212.05,
          "first_token_was_thinking": false,
          "wall_s": 19.0325,
          "eval_count": 256,
          "eval_duration_ns": 18818905000,
          "decode_tok_s": 13.603,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2550.372,
          "load_duration_ns": 858510,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 2
        },
        {
          "num_thread": 6,
          "ttft_ms_CACHED_PREFIX": 211.99,
          "first_token_was_thinking": false,
          "wall_s": 19.0343,
          "eval_count": 256,
          "eval_duration_ns": 18820873000,
          "decode_tok_s": 13.602,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2553.383,
          "load_duration_ns": 931247,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 3
        },
        {
          "num_thread": 6,
          "ttft_ms_CACHED_PREFIX": 212.28,
          "first_token_was_thinking": false,
          "wall_s": 19.0306,
          "eval_count": 256,
          "eval_duration_ns": 18816898000,
          "decode_tok_s": 13.605,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2550.727,
          "load_duration_ns": 856814,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 4
        },
        {
          "num_thread": 6,
          "ttft_ms_CACHED_PREFIX": 212.48,
          "first_token_was_thinking": false,
          "wall_s": 19.0315,
          "eval_count": 256,
          "eval_duration_ns": 18817403000,
          "decode_tok_s": 13.604,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2548.82,
          "load_duration_ns": 1029240,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 5
        }
      ],
      "errors": [],
      "quantization": "Q4_K_M",
      "param_size": "32.9B",
      "context_length": 1048576,
      "think_mode": false,
      "think_note": "think:false accepted",
      "warmup_discarded": {
        "num_thread": 6,
        "ttft_ms_CACHED_PREFIX": 209.37,
        "first_token_was_thinking": false,
        "wall_s": 18.2275,
        "eval_count": 256,
        "eval_duration_ns": 18016446000,
        "decode_tok_s": 14.209,
        "prompt_eval_count": 532,
        "prefill_tok_s_CACHE_ARTIFACT": 2585.914,
        "load_duration_ns": 819227,
        "streamed_chunks": 245,
        "response_chars": 1029,
        "thinking_chars": 0,
        "done_reason": "length"
      },
      "ps_after_load": [
        {
          "name": "nemotron-3.5-lightning:30b-a3b",
          "model": "nemotron-3.5-lightning:30b-a3b",
          "size": 25609340188,
          "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2",
          "details": {
            "parent_model": "",
            "format": "gguf",
            "family": "nemotron_h_moe",
            "families": [
              "nemotron_h_moe"
            ],
            "parameter_size": "32.9B",
            "quantization_level": "Q4_K_M"
          },
          "expires_at": "2026-08-25T19:27:23.825041582Z",
          "size_vram": 0,
          "context_length": 4096
        }
      ],
      "size_vram": 0,
      "size_total": 25609340188,
      "cpu_only_receipt": true,
      "peak_runner_rss_gib": 24.17,
      "peak_runner_threads": 30,
      "loadavg_during_runs": {
        "min": 4.83,
        "median": 5.65,
        "max": 6.48,
        "mean": 5.83,
        "samples": 187,
        "nproc": 16
      },
      "contention_suspected": false,
      "nvidia_after": "no nvidia-smi (no GPU on this box)",
      "derived": {
        "decode_tok_s": {
          "min": 13.602,
          "median": 13.604,
          "max": 13.631,
          "mean": 13.609,
          "n": 5,
          "spread_pct": 0.21
        },
        "wall_s": {
          "min": 18.9909,
          "median": 19.0315,
          "max": 19.0343,
          "mean": 19.024,
          "n": 5,
          "spread_pct": 0.23
        },
        "eval_count": {
          "min": 256,
          "median": 256,
          "max": 256,
          "mean": 256.0,
          "n": 5,
          "spread_pct": 0.0
        },
        "thinking_chars_total": 0
      },
      "finished_utc": "2026-08-25T19:18:59Z"
    },
    {
      "model_id": "m6",
      "model_tag": "nemotron-3.5-lightning:30b-a3b",
      "box_class": "cpu-mini",
      "arm_label": "sweep t=8",
      "num_thread": 8,
      "started_utc": "2026-08-25T19:18:59Z",
      "nproc": 16,
      "num_predict": 256,
      "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c",
      "loadavg_before_arm": "6.32 4.90 5.19 1/324 96403",
      "nvidia_before": "no nvidia-smi (no GPU on this box)",
      "ollama_version": "0.32.15",
      "runs": [
        {
          "num_thread": 8,
          "ttft_ms_CACHED_PREFIX": 182.23,
          "first_token_was_thinking": false,
          "wall_s": 19.3212,
          "eval_count": 256,
          "eval_duration_ns": 19137372000,
          "decode_tok_s": 13.377,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2976.773,
          "load_duration_ns": 583410,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 1
        },
        {
          "num_thread": 8,
          "ttft_ms_CACHED_PREFIX": 192.19,
          "first_token_was_thinking": false,
          "wall_s": 22.1099,
          "eval_count": 256,
          "eval_duration_ns": 21915930000,
          "decode_tok_s": 11.681,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2819.858,
          "load_duration_ns": 923249,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 2
        },
        {
          "num_thread": 8,
          "ttft_ms_CACHED_PREFIX": 190.56,
          "first_token_was_thinking": false,
          "wall_s": 20.9713,
          "eval_count": 256,
          "eval_duration_ns": 20779073000,
          "decode_tok_s": 12.32,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2845.97,
          "load_duration_ns": 891439,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 3
        },
        {
          "num_thread": 8,
          "ttft_ms_CACHED_PREFIX": 191.48,
          "first_token_was_thinking": false,
          "wall_s": 20.6345,
          "eval_count": 256,
          "eval_duration_ns": 20441356000,
          "decode_tok_s": 12.524,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2832.937,
          "load_duration_ns": 956888,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 4
        },
        {
          "num_thread": 8,
          "ttft_ms_CACHED_PREFIX": 189.89,
          "first_token_was_thinking": false,
          "wall_s": 19.7566,
          "eval_count": 256,
          "eval_duration_ns": 19565187000,
          "decode_tok_s": 13.084,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 2855.533,
          "load_duration_ns": 896291,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 5
        }
      ],
      "errors": [],
      "quantization": "Q4_K_M",
      "param_size": "32.9B",
      "context_length": 1048576,
      "think_mode": false,
      "think_note": "think:false accepted",
      "warmup_discarded": {
        "num_thread": 8,
        "ttft_ms_CACHED_PREFIX": 181.41,
        "first_token_was_thinking": false,
        "wall_s": 19.522,
        "eval_count": 256,
        "eval_duration_ns": 19338897000,
        "decode_tok_s": 13.238,
        "prompt_eval_count": 532,
        "prefill_tok_s_CACHE_ARTIFACT": 2990.932,
        "load_duration_ns": 825295,
        "streamed_chunks": 245,
        "response_chars": 1029,
        "thinking_chars": 0,
        "done_reason": "length"
      },
      "ps_after_load": [
        {
          "name": "nemotron-3.5-lightning:30b-a3b",
          "model": "nemotron-3.5-lightning:30b-a3b",
          "size": 25609340188,
          "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2",
          "details": {
            "parent_model": "",
            "format": "gguf",
            "family": "nemotron_h_moe",
            "families": [
              "nemotron_h_moe"
            ],
            "parameter_size": "32.9B",
            "quantization_level": "Q4_K_M"
          },
          "expires_at": "2026-08-25T19:29:44.56160733Z",
          "size_vram": 0,
          "context_length": 4096
        }
      ],
      "size_vram": 0,
      "size_total": 25609340188,
      "cpu_only_receipt": true,
      "peak_runner_rss_gib": 24.17,
      "peak_runner_threads": 34,
      "loadavg_during_runs": {
        "min": 7.64,
        "median": 9.92,
        "max": 11.22,
        "mean": 9.77,
        "samples": 202,
        "nproc": 16
      },
      "contention_suspected": false,
      "nvidia_after": "no nvidia-smi (no GPU on this box)",
      "derived": {
        "decode_tok_s": {
          "min": 11.681,
          "median": 12.524,
          "max": 13.377,
          "mean": 12.597,
          "n": 5,
          "spread_pct": 13.54
        },
        "wall_s": {
          "min": 19.3212,
          "median": 20.6345,
          "max": 22.1099,
          "mean": 20.559,
          "n": 5,
          "spread_pct": 13.51
        },
        "eval_count": {
          "min": 256,
          "median": 256,
          "max": 256,
          "mean": 256.0,
          "n": 5,
          "spread_pct": 0.0
        },
        "thinking_chars_total": 0
      },
      "finished_utc": "2026-08-25T19:21:27Z"
    },
    {
      "model_id": "m6",
      "model_tag": "nemotron-3.5-lightning:30b-a3b",
      "box_class": "cpu-mini",
      "arm_label": "sweep t=12",
      "num_thread": 12,
      "started_utc": "2026-08-25T19:21:28Z",
      "nproc": 16,
      "num_predict": 256,
      "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c",
      "loadavg_before_arm": "9.44 6.94 5.92 1/322 109565",
      "nvidia_before": "no nvidia-smi (no GPU on this box)",
      "ollama_version": "0.32.15",
      "runs": [
        {
          "num_thread": 12,
          "ttft_ms_CACHED_PREFIX": 328.07,
          "first_token_was_thinking": false,
          "wall_s": 35.7783,
          "eval_count": 256,
          "eval_duration_ns": 35448661000,
          "decode_tok_s": 7.222,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1638.9,
          "load_duration_ns": 856671,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 1
        },
        {
          "num_thread": 12,
          "ttft_ms_CACHED_PREFIX": 328.62,
          "first_token_was_thinking": false,
          "wall_s": 35.8708,
          "eval_count": 256,
          "eval_duration_ns": 35540846000,
          "decode_tok_s": 7.203,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1635.444,
          "load_duration_ns": 829012,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 2
        },
        {
          "num_thread": 12,
          "ttft_ms_CACHED_PREFIX": 297.23,
          "first_token_was_thinking": false,
          "wall_s": 35.4097,
          "eval_count": 256,
          "eval_duration_ns": 35112042000,
          "decode_tok_s": 7.291,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1813.799,
          "load_duration_ns": 842510,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 3
        },
        {
          "num_thread": 12,
          "ttft_ms_CACHED_PREFIX": 339.78,
          "first_token_was_thinking": false,
          "wall_s": 35.1568,
          "eval_count": 256,
          "eval_duration_ns": 34815879000,
          "decode_tok_s": 7.353,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1577.554,
          "load_duration_ns": 559461,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 4
        },
        {
          "num_thread": 12,
          "ttft_ms_CACHED_PREFIX": 355.05,
          "first_token_was_thinking": false,
          "wall_s": 35.5212,
          "eval_count": 256,
          "eval_duration_ns": 35164789000,
          "decode_tok_s": 7.28,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1510.96,
          "load_duration_ns": 570619,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 5
        }
      ],
      "errors": [],
      "quantization": "Q4_K_M",
      "param_size": "32.9B",
      "context_length": 1048576,
      "think_mode": false,
      "think_note": "think:false accepted",
      "warmup_discarded": {
        "num_thread": 12,
        "ttft_ms_CACHED_PREFIX": 315.54,
        "first_token_was_thinking": false,
        "wall_s": 36.4019,
        "eval_count": 256,
        "eval_duration_ns": 36084950000,
        "decode_tok_s": 7.094,
        "prompt_eval_count": 532,
        "prefill_tok_s_CACHE_ARTIFACT": 1704.314,
        "load_duration_ns": 792698,
        "streamed_chunks": 245,
        "response_chars": 1029,
        "thinking_chars": 0,
        "done_reason": "length"
      },
      "ps_after_load": [
        {
          "name": "nemotron-3.5-lightning:30b-a3b",
          "model": "nemotron-3.5-lightning:30b-a3b",
          "size": 25609340188,
          "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2",
          "details": {
            "parent_model": "",
            "format": "gguf",
            "family": "nemotron_h_moe",
            "families": [
              "nemotron_h_moe"
            ],
            "parameter_size": "32.9B",
            "quantization_level": "Q4_K_M"
          },
          "expires_at": "2026-08-25T19:32:27.706964501Z",
          "size_vram": 0,
          "context_length": 4096
        }
      ],
      "size_vram": 0,
      "size_total": 25609340188,
      "cpu_only_receipt": true,
      "peak_runner_rss_gib": 24.17,
      "peak_runner_threads": 42,
      "loadavg_during_runs": {
        "min": 12.22,
        "median": 16.41,
        "max": 17.48,
        "mean": 16.1,
        "samples": 349,
        "nproc": 16
      },
      "contention_suspected": false,
      "nvidia_after": "no nvidia-smi (no GPU on this box)",
      "derived": {
        "decode_tok_s": {
          "min": 7.203,
          "median": 7.28,
          "max": 7.353,
          "mean": 7.27,
          "n": 5,
          "spread_pct": 2.06
        },
        "wall_s": {
          "min": 35.1568,
          "median": 35.5212,
          "max": 35.8708,
          "mean": 35.547,
          "n": 5,
          "spread_pct": 2.01
        },
        "eval_count": {
          "min": 256,
          "median": 256,
          "max": 256,
          "mean": 256.0,
          "n": 5,
          "spread_pct": 0.0
        },
        "thinking_chars_total": 0
      },
      "finished_utc": "2026-08-25T19:25:25Z"
    },
    {
      "model_id": "m6",
      "model_tag": "nemotron-3.5-lightning:30b-a3b",
      "box_class": "cpu-mini",
      "arm_label": "sweep t=16",
      "num_thread": 16,
      "started_utc": "2026-08-25T19:25:26Z",
      "nproc": 16,
      "num_predict": 256,
      "prompt_sha256": "90eedd0c53f9554ae3837674504fcb7090013d9a432a653352183c0f25a7ce5c",
      "loadavg_before_arm": "17.04 12.34 8.33 1/327 130183",
      "nvidia_before": "no nvidia-smi (no GPU on this box)",
      "ollama_version": "0.32.15",
      "runs": [
        {
          "num_thread": 16,
          "ttft_ms_CACHED_PREFIX": 311.95,
          "first_token_was_thinking": false,
          "wall_s": 183.1402,
          "eval_count": 256,
          "eval_duration_ns": 182826868000,
          "decode_tok_s": 1.4,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1756.535,
          "load_duration_ns": 1785665,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 1
        },
        {
          "num_thread": 16,
          "ttft_ms_CACHED_PREFIX": 1616.03,
          "first_token_was_thinking": false,
          "wall_s": 202.6561,
          "eval_count": 256,
          "eval_duration_ns": 201038914000,
          "decode_tok_s": 1.273,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 329.927,
          "load_duration_ns": 556276,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 2
        },
        {
          "num_thread": 16,
          "ttft_ms_CACHED_PREFIX": 1633.89,
          "first_token_was_thinking": false,
          "wall_s": 180.4327,
          "eval_count": 256,
          "eval_duration_ns": 178795926000,
          "decode_tok_s": 1.432,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 326.397,
          "load_duration_ns": 796014,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 3
        },
        {
          "num_thread": 16,
          "ttft_ms_CACHED_PREFIX": 392.96,
          "first_token_was_thinking": false,
          "wall_s": 206.4299,
          "eval_count": 256,
          "eval_duration_ns": 206035925000,
          "decode_tok_s": 1.243,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 1386.272,
          "load_duration_ns": 2520230,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 4
        },
        {
          "num_thread": 16,
          "ttft_ms_CACHED_PREFIX": 1528.65,
          "first_token_was_thinking": false,
          "wall_s": 208.384,
          "eval_count": 256,
          "eval_duration_ns": 206849937000,
          "decode_tok_s": 1.238,
          "prompt_eval_count": 532,
          "prefill_tok_s_CACHE_ARTIFACT": 348.79,
          "load_duration_ns": 810162,
          "streamed_chunks": 245,
          "response_chars": 1029,
          "thinking_chars": 0,
          "done_reason": "length",
          "run": 5
        }
      ],
      "errors": [],
      "quantization": "Q4_K_M",
      "param_size": "32.9B",
      "context_length": 1048576,
      "think_mode": false,
      "think_note": "think:false accepted",
      "warmup_discarded": {
        "num_thread": 16,
        "ttft_ms_CACHED_PREFIX": 1406.93,
        "first_token_was_thinking": false,
        "wall_s": 209.8942,
        "eval_count": 256,
        "eval_duration_ns": 208485765000,
        "decode_tok_s": 1.228,
        "prompt_eval_count": 532,
        "prefill_tok_s_CACHE_ARTIFACT": 379.169,
        "load_duration_ns": 779063,
        "streamed_chunks": 245,
        "response_chars": 1029,
        "thinking_chars": 0,
        "done_reason": "length"
      },
      "ps_after_load": [
        {
          "name": "nemotron-3.5-lightning:30b-a3b",
          "model": "nemotron-3.5-lightning:30b-a3b",
          "size": 25609340188,
          "digest": "e7a64ff15fb174c42b4f463e5c888c4f2c7b9cabf9e8d65a1c0874405426c1b2",
          "details": {
            "parent_model": "",
            "format": "gguf",
            "family": "nemotron_h_moe",
            "families": [
              "nemotron_h_moe"
            ],
            "parameter_size": "32.9B",
            "quantization_level": "Q4_K_M"
          },
          "expires_at": "2026-08-25T19:39:17.763712741Z",
          "size_vram": 0,
          "context_length": 4096
        }
      ],
      "size_vram": 0,
      "size_total": 25609340188,
      "cpu_only_receipt": true,
      "peak_runner_rss_gib": 24.17,
      "peak_runner_threads": 50,
      "loadavg_during_runs": {
        "min": 16.88,
        "median": 19.02,
        "max": 21.94,
        "mean": 19.13,
        "samples": 1933,
        "nproc": 16
      },
      "contention_suspected": false,
      "nvidia_after": "no nvidia-smi (no GPU on this box)",
      "derived": {
        "decode_tok_s": {
          "min": 1.238,
          "median": 1.273,
          "max": 1.432,
          "mean": 1.317,
          "n": 5,
          "spread_pct": 15.24
        },
        "wall_s": {
          "min": 180.4327,
          "median": 202.6561,
          "max": 208.384,
          "mean": 196.209,
          "n": 5,
          "spread_pct": 13.79
        },
        "eval_count": {
          "min": 256,
          "median": 256,
          "max": 256,
          "mean": 256.0,
          "n": 5,
          "spread_pct": 0.0
        },
        "thinking_chars_total": 0
      },
      "finished_utc": "2026-08-25T19:45:39Z"
    }
  ],
  "finished_utc": "2026-08-25T19:45:39Z"
}