{
  "base_url": "http://127.0.0.1:39268/v1",
  "completed": 8,
  "concurrency": 4,
  "context_distribution": {
    "4096": 8
  },
  "context_seed": 0,
  "context_tokens": 4096,
  "engine": "sglang-5f55db35-dflash2-k12",
  "failed": 0,
  "failures": [],
  "finished_at": "2026-09-05T03:14:11Z",
  "gpu": "RTX PRO 6000 Blackwell Max-Q TP1",
  "manifest": null,
  "max_context_tokens": 262144,
  "max_tokens": 512,
  "measurement_protocol": "capacity-v3",
  "metrics": {
    "content_chunks": 120,
    "content_chunks_s": null,
    "decode_tok_s_p50": 85.96744283492825,
    "decode_tok_s_p95": 109.77358611876791,
    "e2e_p50_ms": 722.554299980402,
    "e2e_p95_ms": 1073.5974999843165,
    "effective_prefill_tok_s_p50": 12606.91715938463,
    "effective_prefill_tok_s_p95": 26819.084389495554,
    "generation_p50_ms": 511.160199996084,
    "generation_p95_ms": 778.7924000294879,
    "mean_inter_token_latency_ms_p50": 10.64917083325175,
    "mean_inter_token_latency_ms_p95": 16.224841667280998,
    "output_token_sources": {
      "usage": 8
    },
    "output_tokens": 389,
    "prefix_cache_hit_avg": null,
    "prompt_token_samples": 8,
    "prompt_tokens": 28904,
    "prompt_tokens_p50": 3613.0,
    "prompt_tokens_p95": 3613.0,
    "reasoning_chunks": 0,
    "throughput_tok_s": 226.05421894700274,
    "time_to_first_output_p50_ms": 247.51239991746843,
    "time_to_first_output_p95_ms": 294.8050999548286,
    "ttft_p50_ms": 247.51299992203712,
    "ttft_p95_ms": 294.8052999563515,
    "ttft_samples": 8
  },
  "model": "qwen38-27b-inferact-nvfp4-pro6000-dflash2-k12",
  "request_timings": [
    {
      "content_chunks": 15,
      "decode_tok_s": 84.55624267069702,
      "e2e_ms": 854.2582000372931,
      "effective_prefill_tok_s": 12606.91715938463,
      "generation_ms": 567.6695000147447,
      "mean_inter_token_latency_ms": 11.826447916973848,
      "output_token_source": "usage",
      "output_tokens": 49,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 0,
      "time_to_first_output_ms": 286.58870002254844,
      "ttft_ms": 286.58950002864003,
      "visible_generation_ms": 567.6687000086531
    },
    {
      "content_chunks": 17,
      "decode_tok_s": 61.63388343052981,
      "e2e_ms": 1073.5974999843165,
      "effective_prefill_tok_s": 12255.554603884399,
      "generation_ms": 778.7924000294879,
      "mean_inter_token_latency_ms": 16.224841667280998,
      "output_token_source": "usage",
      "output_tokens": 49,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 1,
      "time_to_first_output_ms": 294.8050999548286,
      "ttft_ms": 294.8052999563515,
      "visible_generation_ms": 778.792200027965
    },
    {
      "content_chunks": 14,
      "decode_tok_s": 109.77358611876791,
      "e2e_ms": 722.554299980402,
      "effective_prefill_tok_s": 12272.405976611015,
      "generation_ms": 428.1540000811219,
      "mean_inter_token_latency_ms": 9.109659576194083,
      "output_token_source": "usage",
      "output_tokens": 48,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 2,
      "time_to_first_output_ms": 294.40029989928007,
      "ttft_ms": 294.40049990080297,
      "visible_generation_ms": 428.153800079599
    },
    {
      "content_chunks": 14,
      "decode_tok_s": 104.1598176164682,
      "e2e_ms": 754.8728999681771,
      "effective_prefill_tok_s": 12287.335237731639,
      "generation_ms": 460.83029999863356,
      "mean_inter_token_latency_ms": 9.600631249971533,
      "output_token_source": "usage",
      "output_tokens": 49,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 3,
      "time_to_first_output_ms": 294.0425999695435,
      "ttft_ms": 294.04289997182786,
      "visible_generation_ms": 460.8299999963492
    },
    {
      "content_chunks": 15,
      "decode_tok_s": 76.30150092886755,
      "e2e_ms": 833.7861000327393,
      "effective_prefill_tok_s": 17649.97840570588,
      "generation_ms": 629.0833000093699,
      "mean_inter_token_latency_ms": 13.10590208352854,
      "output_token_source": "usage",
      "output_tokens": 49,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 4,
      "time_to_first_output_ms": 204.7028000233695,
      "ttft_ms": 204.70330002717674,
      "visible_generation_ms": 629.0828000055626
    },
    {
      "content_chunks": 13,
      "decode_tok_s": 102.34382602448323,
      "e2e_ms": 706.7487000022084,
      "effective_prefill_tok_s": 14597.24846595457,
      "generation_ms": 459.2363000847399,
      "mean_inter_token_latency_ms": 9.770985108185956,
      "output_token_source": "usage",
      "output_tokens": 48,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 5,
      "time_to_first_output_ms": 247.51239991746843,
      "ttft_ms": 247.51299992203712,
      "visible_generation_ms": 459.2357000801712
    },
    {
      "content_chunks": 16,
      "decode_tok_s": 85.96744283492825,
      "e2e_ms": 702.6700000278652,
      "effective_prefill_tok_s": 23167.473978634825,
      "generation_ms": 546.7186000896618,
      "mean_inter_token_latency_ms": 11.632310640205569,
      "output_token_source": "usage",
      "output_tokens": 48,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 6,
      "time_to_first_output_ms": 155.9513999382034,
      "ttft_ms": 155.9515999397263,
      "visible_generation_ms": 546.7184000881389
    },
    {
      "content_chunks": 16,
      "decode_tok_s": 93.90402460983412,
      "e2e_ms": 645.8776999497786,
      "effective_prefill_tok_s": 26819.084389495554,
      "generation_ms": 511.160199996084,
      "mean_inter_token_latency_ms": 10.64917083325175,
      "output_token_source": "usage",
      "output_tokens": 49,
      "planned_context_tokens": 4096,
      "prompt_tokens": 3613,
      "reasoning_chunks": 0,
      "request_index": 7,
      "time_to_first_output_ms": 134.71749995369464,
      "ttft_ms": 134.7175999544561,
      "visible_generation_ms": 511.1600999953225
    }
  ],
  "requests": 8,
  "run_id": "benchmark-20260905T031409Z",
  "schema": "anvil-serving.benchmark/v1",
  "serve_flags": {
    "no_thinking": true,
    "reasoning_effort": null,
    "shared_prefix_burst": false,
    "thinking_mode": "disabled"
  },
  "source_recipe": null,
  "started_at": "2026-09-05T03:14:09Z",
  "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": 1720.8261000923812
}