{
  "base_url": "http://127.0.0.1:39050/v1",
  "completed": 11,
  "concurrency": 1,
  "context_distribution": {
    "29696": 12
  },
  "context_seed": 0,
  "context_tokens": 32768,
  "engine": "sglang-v0.5.16-wsl2-nomultimem",
  "failed": 1,
  "failures": [
    {
      "error_type": "ValueError",
      "request_index": 7
    }
  ],
  "finished_at": "2026-08-01T14:25:47Z",
  "gpu": "dual-rtx-pro-6000-blackwell-max-q-tp2-pcie",
  "manifest": null,
  "max_context_tokens": 32768,
  "max_tokens": 2048,
  "measurement_protocol": "capacity-v4-reasoning",
  "metrics": {
    "content_chunks": 380,
    "content_chunks_s": null,
    "decode_tok_s_p50": 11.529481100839698,
    "decode_tok_s_p95": 11.675242541970967,
    "e2e_p50_ms": 32116.06490000122,
    "e2e_p95_ms": 95481.33920000328,
    "effective_prefill_tok_s_p50": 7817.931879370884,
    "effective_prefill_tok_s_p95": 7857.495512965675,
    "generation_p50_ms": 29402.88440000222,
    "generation_p95_ms": 92778.36960000059,
    "mean_inter_token_latency_ms_p50": 86.73417227139298,
    "mean_inter_token_latency_ms_p95": 87.71314571427605,
    "output_token_sources": {
      "usage": 11
    },
    "output_tokens": 4650,
    "prefix_cache_hit_avg": null,
    "prompt_token_samples": 11,
    "prompt_tokens": 232582,
    "prompt_tokens_p50": 21144.0,
    "prompt_tokens_p95": 21147.0,
    "reasoning_chunks": 4248,
    "throughput_tok_s": 7.59630572878913,
    "time_to_first_output_p50_ms": 2704.699099998834,
    "time_to_first_output_p95_ms": 2982.1330000013404,
    "ttft_p50_ms": 29105.986899998243,
    "ttft_p95_ms": 92102.9097999999,
    "ttft_samples": 11
  },
  "model": "deepseek-v4-flash-0731-tp2",
  "request_timings": [
    {
      "content_chunks": 27,
      "decode_tok_s": 11.400799639058453,
      "e2e_ms": 15261.973399999988,
      "effective_prefill_tok_s": 7090.8976896706135,
      "generation_ms": 12279.840399998648,
      "mean_inter_token_latency_ms": 87.71314571427605,
      "output_token_source": "usage",
      "output_tokens": 141,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21146,
      "reasoning_chunks": 112,
      "request_index": 0,
      "time_to_first_output_ms": 2982.1330000013404,
      "ttft_ms": 12863.746200000605,
      "visible_generation_ms": 2398.2271999993827
    },
    {
      "content_chunks": 33,
      "decode_tok_s": 11.641889648254045,
      "e2e_ms": 60168.20430000007,
      "effective_prefill_tok_s": 7822.642671955016,
      "generation_ms": 57464.89790000123,
      "mean_inter_token_latency_ms": 85.89670837070439,
      "output_token_source": "usage",
      "output_tokens": 670,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21147,
      "reasoning_chunks": 635,
      "request_index": 1,
      "time_to_first_output_ms": 2703.306399998837,
      "ttft_ms": 57421.694799999386,
      "visible_generation_ms": 2746.509500000684
    },
    {
      "content_chunks": 40,
      "decode_tok_s": 11.444208271590417,
      "e2e_ms": 26733.989999996993,
      "effective_prefill_tok_s": 7819.202409559506,
      "generation_ms": 24029.62209999896,
      "mean_inter_token_latency_ms": 87.38044399999622,
      "output_token_source": "usage",
      "output_tokens": 276,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21146,
      "reasoning_chunks": 234,
      "request_index": 2,
      "time_to_first_output_ms": 2704.367899998033,
      "ttft_ms": 23195.865299996512,
      "visible_generation_ms": 3538.1247000004805
    },
    {
      "content_chunks": 35,
      "decode_tok_s": 11.62013301767498,
      "e2e_ms": 22067.75270000071,
      "effective_prefill_tok_s": 7817.931879370884,
      "generation_ms": 19362.945300003048,
      "mean_inter_token_latency_ms": 86.05753466668021,
      "output_token_source": "usage",
      "output_tokens": 226,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21146,
      "reasoning_chunks": 189,
      "request_index": 3,
      "time_to_first_output_ms": 2704.8073999976623,
      "ttft_ms": 19114.38249999992,
      "visible_generation_ms": 2953.3702000007906
    },
    {
      "content_chunks": 34,
      "decode_tok_s": 11.529481100839698,
      "e2e_ms": 32116.06490000122,
      "effective_prefill_tok_s": 7793.067951066215,
      "generation_ms": 29402.88440000222,
      "mean_inter_token_latency_ms": 86.73417227139298,
      "output_token_source": "usage",
      "output_tokens": 340,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21144,
      "reasoning_chunks": 304,
      "request_index": 4,
      "time_to_first_output_ms": 2713.180499998998,
      "ttft_ms": 29105.986899998243,
      "visible_generation_ms": 3010.078000002977
    },
    {
      "content_chunks": 39,
      "decode_tok_s": 11.468190318360513,
      "e2e_ms": 95481.33920000328,
      "effective_prefill_tok_s": 7822.5075117304195,
      "generation_ms": 92778.36960000059,
      "mean_inter_token_latency_ms": 87.19771578947423,
      "output_token_source": "usage",
      "output_tokens": 1065,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21144,
      "reasoning_chunks": 1024,
      "request_index": 5,
      "time_to_first_output_ms": 2702.969600002689,
      "ttft_ms": 92102.9097999999,
      "visible_generation_ms": 3378.4294000033697
    },
    {
      "content_chunks": 39,
      "decode_tok_s": 11.675242541970967,
      "e2e_ms": 36876.14759999997,
      "effective_prefill_tok_s": 7827.062890859275,
      "generation_ms": 34174.878900001204,
      "mean_inter_token_latency_ms": 85.65132556391279,
      "output_token_source": "usage",
      "output_tokens": 400,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21143,
      "reasoning_chunks": 359,
      "request_index": 6,
      "time_to_first_output_ms": 2701.2686999987636,
      "ttft_ms": 33610.28470000019,
      "visible_generation_ms": 3265.8628999997745
    },
    {
      "content_chunks": 35,
      "decode_tok_s": 11.49579151845527,
      "e2e_ms": 22263.1865000003,
      "effective_prefill_tok_s": 7857.495512965675,
      "generation_ms": 19572.38000000143,
      "mean_inter_token_latency_ms": 86.98835555556191,
      "output_token_source": "usage",
      "output_tokens": 226,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21143,
      "reasoning_chunks": 189,
      "request_index": 8,
      "time_to_first_output_ms": 2690.806499998871,
      "ttft_ms": 19174.827200000436,
      "visible_generation_ms": 3088.3592999998655
    },
    {
      "content_chunks": 28,
      "decode_tok_s": 11.619032846741995,
      "e2e_ms": 23188.331699999253,
      "effective_prefill_tok_s": 7816.3962860079755,
      "generation_ms": 20483.63260000042,
      "mean_inter_token_latency_ms": 86.06568319327907,
      "output_token_source": "usage",
      "output_tokens": 239,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21141,
      "reasoning_chunks": 209,
      "request_index": 9,
      "time_to_first_output_ms": 2704.699099998834,
      "ttft_ms": 20805.97530000159,
      "visible_generation_ms": 2382.3563999976614
    },
    {
      "content_chunks": 31,
      "decode_tok_s": 11.463574593604438,
      "e2e_ms": 51295.652699998755,
      "effective_prefill_tok_s": 7809.841930220966,
      "generation_ms": 48588.68369999982,
      "mean_inter_token_latency_ms": 87.2328253141828,
      "output_token_source": "usage",
      "output_tokens": 558,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21141,
      "reasoning_chunks": 525,
      "request_index": 10,
      "time_to_first_output_ms": 2706.968999998935,
      "ttft_ms": 48695.597400001134,
      "visible_generation_ms": 2600.0552999976208
    },
    {
      "content_chunks": 39,
      "decode_tok_s": 11.5784143966843,
      "e2e_ms": 46579.886100000294,
      "effective_prefill_tok_s": 7815.124344982451,
      "generation_ms": 43874.74680000014,
      "mean_inter_token_latency_ms": 86.36761181102389,
      "output_token_source": "usage",
      "output_tokens": 509,
      "planned_context_tokens": 29696,
      "prompt_tokens": 21141,
      "reasoning_chunks": 468,
      "request_index": 11,
      "time_to_first_output_ms": 2705.1393000001553,
      "ttft_ms": 43209.467500000756,
      "visible_generation_ms": 3370.418599999539
    }
  ],
  "requests": 12,
  "run_id": "benchmark-20260801T141535Z",
  "schema": "anvil-serving.benchmark/v1",
  "serve_flags": {
    "no_thinking": false,
    "reasoning_effort": "low",
    "shared_prefix_burst": false,
    "thinking_mode": "default"
  },
  "source_recipe": null,
  "started_at": "2026-08-01T14:15:35Z",
  "tier": null,
  "timing_methodology": {
    "clock": "client time.perf_counter",
    "decode": "usage completion tokens after the first token divided by client-observed generation time; completion tokens may include reasoning tokens",
    "effective_prefill": "usage.prompt_tokens divided by client-observed time to first output; includes queueing, scheduling, prefill, and first-token work",
    "generation": "client-observed E2E minus time to first output",
    "mean_inter_token_latency": "client-observed generation time divided by completion tokens after the first token; not a raw per-token timestamp trace",
    "time_to_first_output": "request start through the first non-empty streamed reasoning or content delta",
    "ttft": "request start through the first non-empty streamed content delta"
  },
  "wall_clock_ms": 612139.659200002
}
