{
  "context": {
    "cap_tokens": 126464,
    "max_model_len": 131072,
    "targets": [
      {
        "attempted_context_tokens": 32768,
        "chars_per_token": 3.0,
        "clamped_tokens": 32768,
        "content_chunks": 8,
        "decode_tok_s": 102.09137001049544,
        "e2e_ms": 19559.208200000285,
        "effective_prefill_tok_s": 1573.8762770383516,
        "estimated_prompt_tokens": 32768,
        "generation_ms": 4417.611400000169,
        "mean_inter_token_latency_ms": 9.79514722838175,
        "output_token_source": "usage",
        "output_tokens": 452,
        "prompt_tokens": 23831,
        "reasoning_chunks": 118,
        "status": "passed",
        "target_tokens": 32768,
        "time_to_first_output_ms": 15141.596800000116,
        "ttft_ms": 19301.395099995716,
        "usage": {
          "completion_tokens": 452,
          "prompt_tokens": 23831,
          "prompt_tokens_details": {
            "cached_tokens": 0
          },
          "total_tokens": 24283
        },
        "visible_generation_ms": 257.8131000045687
      },
      {
        "attempted_context_tokens": 65536,
        "chars_per_token": 4.125,
        "clamped_tokens": 65536,
        "content_chunks": 13,
        "decode_tok_s": 88.95471980233219,
        "e2e_ms": 42114.74360000284,
        "effective_prefill_tok_s": 1746.6790912382821,
        "estimated_prompt_tokens": 65535,
        "generation_ms": 3518.644100004167,
        "mean_inter_token_latency_ms": 11.24167444090788,
        "output_token_source": "usage",
        "output_tokens": 314,
        "prompt_tokens": 67415,
        "reasoning_chunks": 84,
        "status": "passed",
        "target_tokens": 65536,
        "time_to_first_output_ms": 38596.09949999867,
        "ttft_ms": 41630.009900000005,
        "usage": {
          "completion_tokens": 314,
          "prompt_tokens": 67415,
          "prompt_tokens_details": {
            "cached_tokens": 5632
          },
          "total_tokens": 67729
        },
        "visible_generation_ms": 484.73370000283467
      },
      {
        "attempted_context_tokens": 126464,
        "chars_per_token": 4.01,
        "clamped_tokens": 126464,
        "content_chunks": 5,
        "decode_tok_s": 113.3599706364023,
        "e2e_ms": 85910.35009999905,
        "effective_prefill_tok_s": 1484.2602819928904,
        "estimated_prompt_tokens": 126463,
        "generation_ms": 1164.4321999992826,
        "mean_inter_token_latency_ms": 8.821456060600626,
        "output_token_source": "usage",
        "output_tokens": 133,
        "prompt_tokens": 125785,
        "reasoning_chunks": 31,
        "status": "passed",
        "target_tokens": 128000,
        "time_to_first_output_ms": 84745.91789999977,
        "ttft_ms": 85746.72709999868,
        "usage": {
          "completion_tokens": 133,
          "prompt_tokens": 125785,
          "prompt_tokens_details": {
            "cached_tokens": 5632
          },
          "total_tokens": 125918
        },
        "visible_generation_ms": 163.62300000037067
      }
    ]
  },
  "evaluation_protocol": {
    "budget_semantics": "visible_answer_tokens plus reasoning_headroom_tokens are sent as one max_tokens cap; the endpoint does not hard-partition the channels",
    "minimum_comparable_repetitions": 3,
    "minimum_pass_rate": 1.0,
    "reasoning_headroom_tokens": 2048,
    "records_finish_reason": true,
    "records_full_visible_answer": true,
    "records_reasoning_channel_metadata": true,
    "repetitions": 1,
    "version": 3,
    "visible_answer_tokens": 512
  },
  "failures": [],
  "identity": {
    "base_url": "http://127.0.0.1:39062/v1",
    "candidate_id": "deepseek-v4-flash-0731-r16-b12x-dspark5-tp2-128k",
    "config_id": "r16-b12x-dspark5-tp2-128k-seqs8",
    "engine": "vllm-r16-b12x-dspark5-sm120",
    "gpu": "dual-rtx-pro-6000-blackwell-max-q-tp2",
    "model": "deepseek-v4-flash-0731-r16-b12x-dspark5-tp2-128k",
    "started_at": "2026-08-02T02:24:42Z"
  },
  "intelligence": {
    "checks": [],
    "status": "not_run"
  },
  "run_id": "fast-bakeoff-20260802T022442Z",
  "schema": "anvil-serving.fast-tier-bakeoff/v1",
  "score_inputs": {
    "e2e_p50_ms": 42114.74360000284,
    "intelligence_pass_rate": null,
    "operational_fit_notes": [
      "endpoint was already loaded; benchmark did not start or stop serves"
    ],
    "session_recall_passed": false,
    "thinking_mode": "default",
    "tool_call_passed": false,
    "ttft_p50_ms": 41630.009900000005,
    "usable_context_tokens": 128000,
    "voice_latency_ms": null
  },
  "selection": {
    "context_targets": [
      32768,
      65536,
      128000
    ],
    "endpoint_already_loaded": true,
    "requests_per_context": 1,
    "suites": [
      "context"
    ]
  },
  "session": {
    "checks": [],
    "status": "not_run"
  },
  "source_recipe": {
    "ref": null,
    "serve_command": null
  },
  "suites": {},
  "thinking": {
    "chat_template_kwargs": null,
    "control_evidence": "docs/findings/2026-08-01-deepseek-v4-flash-0731-nvfp4-evidence/reasoning-r16-b12x-dspark5-low-control.json",
    "control_evidence_sha256": "4e911af139384b6593d0ab9e8d4a4a37a7dd5ebf67d6c33dee4c061bd8e77e46",
    "control_mechanism": "reasoning_effort",
    "control_requested": "low",
    "control_status": "verified",
    "mode": "default",
    "reasoning_effort": "low",
    "unsupported": false
  },
  "timing": {
    "chat": {
      "content_chunks": 26,
      "decode_tok_s_p50": 102.09137001049544,
      "decode_tok_s_p95": 113.3599706364023,
      "e2e_p50_ms": 42114.74360000284,
      "e2e_p95_ms": 85910.35009999905,
      "effective_prefill_tok_s_p50": 1573.8762770383516,
      "effective_prefill_tok_s_p95": 1746.6790912382821,
      "generation_p50_ms": 3518.644100004167,
      "generation_p95_ms": 4417.611400000169,
      "mean_inter_token_latency_ms_p50": 9.79514722838175,
      "mean_inter_token_latency_ms_p95": 11.24167444090788,
      "output_tokens": 899,
      "prompt_token_samples": 3,
      "prompt_tokens": 217031,
      "prompt_tokens_p50": 67415.0,
      "prompt_tokens_p95": 125785.0,
      "reasoning_chunks": 233,
      "time_to_first_output_p50_ms": 38596.09949999867,
      "time_to_first_output_p95_ms": 84745.91789999977,
      "ttft_p50_ms": 41630.009900000005,
      "ttft_p95_ms": 85746.72709999868,
      "ttft_samples": 3
    },
    "wall_ms": 147588.7131999989
  },
  "tool": {
    "checks": [],
    "status": "not_run"
  },
  "voice": {
    "llm_latency_ms": null,
    "status": "not_run",
    "stt_latency_ms": null,
    "total_turn_latency_ms": null,
    "tts_latency_ms": null
  }
}
