{
 "schemaVersion": "callerlab.benchmark/1.0",
 "publishedAt": "2026-10-13T05:09:38+00:00",
 "standards": [
  {
   "version": "0.0.0",
   "status": "proposed",
   "frozenAt": null,
   "manifestSha256": null,
   "harnessVersion": "0.1.0",
   "gitTag": "standard-v0 (at freeze)",
   "judge": {
    "mode": "single"
   }
  }
 ],
 "conditions": [
  {
   "id": "llm-api-iad",
   "layer": "llm",
   "transport": "api",
   "region": "US (us-east, Cloudflare IAD)",
   "audio": "text only (no audio)",
   "description": "12 real phone-agent contexts; streaming chat completions; max_tokens 400; caller persona prompt; sequential. Client: callerlab box, 8 vCPU Linux, single-threaded sequential requests. 4 providers were left out under their terms of service (see Fairness)."
  },
  {
   "id": "tts-api-iad",
   "layer": "tts",
   "transport": "api",
   "region": "US (us-east, Cloudflare IAD)",
   "audio": "8 kHz mu-law (native, or transcoded by ffmpeg where a voice has no 8 kHz output; transcode time excluded from time to first audio)",
   "description": "6 caller lines (money-move consent, a dispute, anger with mild profanity, sarcasm, joking, frustration), 2 repetitions each, streaming synthesis, sequential requests. 1 provider was left out under its terms of service (see Fairness)."
  },
  {
   "id": "stt-api-iad",
   "layer": "stt",
   "transport": "api",
   "region": "US (us-east, Cloudflare IAD)",
   "audio": "8 kHz mu-law clips from the TTS run (clean synthetic speech)",
   "description": "Non-streaming file upload of every clip from the TTS run, including clips from the provider left out of the TTS table; latency = request to full transcript. Not streaming 'final' latency and not noisy real callers."
  },
  {
   "id": "e2e-pstn-gather",
   "layer": "e2e",
   "transport": "pstn-gather",
   "carrier": "Twilio",
   "region": "US",
   "noise": null,
   "quietOnly": true,
   "audio": "8 kHz mu-law (PSTN)",
   "description": "Live PSTN calls over Twilio Gather (half-duplex) to our own test agents on a hosted voice-agent platform; reply gap = end of caller audio to first agent audio, read off the two-channel recording; quiet calls; one measured reply gap per call."
  }
 ],
 "providers": [
  {
   "id": "baseten-llm",
   "kind": "llm",
   "displayName": "Baseten",
   "disclosure": "named",
   "anonLabel": null,
   "tosChecked": true,
   "tosVerdict": "ok",
   "conflict": null,
   "tosCheckedAt": "2026-10-14"
  },
  {
   "id": "groq-llm",
   "kind": "llm",
   "displayName": "Groq",
   "disclosure": "named",
   "anonLabel": null,
   "tosChecked": true,
   "tosVerdict": "ok",
   "conflict": null,
   "tosCheckedAt": "2026-10-14"
  },
  {
   "id": "llm-host-a",
   "kind": "llm",
   "displayName": "Host A",
   "disclosure": "anonymized",
   "anonLabel": "Host A",
   "tosChecked": false,
   "tosVerdict": "unknown",
   "conflict": null
  },
  {
   "id": "tts-vendor-1",
   "kind": "tts",
   "displayName": "TTS 1",
   "disclosure": "anonymized",
   "anonLabel": "TTS 1",
   "tosChecked": false,
   "tosVerdict": "unknown",
   "conflict": null
  },
  {
   "id": "tts-vendor-2",
   "kind": "tts",
   "displayName": "TTS 2",
   "disclosure": "anonymized",
   "anonLabel": "TTS 2",
   "tosChecked": false,
   "tosVerdict": "unknown",
   "conflict": null
  },
  {
   "id": "stt-vendor-1",
   "kind": "stt",
   "displayName": "STT 1",
   "disclosure": "anonymized",
   "anonLabel": "STT 1",
   "tosChecked": false,
   "tosVerdict": "unknown",
   "conflict": null
  },
  {
   "id": "ref-agents",
   "kind": "platform",
   "displayName": "Reference agents (callerlab's own test agents)",
   "disclosure": "named",
   "anonLabel": null,
   "tosChecked": false,
   "tosVerdict": "unknown",
   "conflict": null
  }
 ],
 "runs": [
  {
   "id": "voicebench-llm-2026-10-13",
   "layer": "llm",
   "standardVersion": null,
   "startedAt": "2026-10-13T05:00:59+00:00",
   "conditionId": "llm-api-iad",
   "dataStatus": "real",
   "calls": 0,
   "costUsd": null,
   "source": "voicebench v0.1 caller_next_line (voicebench-results.json, llm-samples.json)"
  },
  {
   "id": "voicebench-tts-2026-10-13",
   "layer": "tts",
   "standardVersion": null,
   "startedAt": "2026-10-13T05:01:12+00:00",
   "conditionId": "tts-api-iad",
   "dataStatus": "real",
   "calls": 0,
   "costUsd": null,
   "source": "voicebench v0.1 tts_caller_line (voicebench-results.json, tts-samples.json)"
  },
  {
   "id": "voicebench-stt-2026-10-13",
   "layer": "stt",
   "standardVersion": null,
   "startedAt": "2026-10-13T05:01:12+00:00",
   "conditionId": "stt-api-iad",
   "dataStatus": "real",
   "calls": 0,
   "costUsd": null,
   "source": "voicebench v0.1 (voicebench-results.json, tts-samples.json stt_latency_s)"
  },
  {
   "id": "e2e-ref-agents-2026-10-12",
   "layer": "e2e",
   "standardVersion": null,
   "startedAt": "2026-10-11T00:00:00-04:00",
   "endedAt": "2026-10-12T23:59:59-04:00",
   "conditionId": "e2e-pstn-gather",
   "dataStatus": "real",
   "calls": 16,
   "costUsd": null,
   "source": "results/runs/*/report/report.json replyGaps.latency p50 (16 runs, Oct 11-12, 2026 ET)",
   "consent": {
    "source": "lab",
    "basis": "own-agent"
   }
  }
 ],
 "results": [
  {
   "id": "llm-baseten-deepseek-ai-deepseek-v4-1-flash",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "baseten-llm",
   "subject": "deepseek-ai/DeepSeek-V4.1-Flash",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": null,
     "n": 12,
     "unit": "usd",
     "sampleState": "not_measured",
     "estimated": true,
     "derived": false,
     "reason": "Not shown: the provider’s list price is being re-checked"
    },
    "persona": {
     "value": 4.916667,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 4.833333,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "llm-baseten-zai-org-glm-5-3-flash",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "baseten-llm",
   "subject": "zai-org/GLM-5.3-Flash",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": null,
     "n": 12,
     "unit": "usd",
     "sampleState": "not_measured",
     "estimated": true,
     "derived": false,
     "reason": "Not shown: the provider’s list price is being re-checked"
    },
    "persona": {
     "value": 4.916667,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "llm-groq-model-3",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "groq-llm",
   "subject": "Model 3",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies · params {\"reasoning_effort\": \"low\"}",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": null,
     "n": 12,
     "unit": "usd",
     "sampleState": "not_measured",
     "estimated": true,
     "derived": false,
     "reason": "Not shown: the provider’s list price is being re-checked"
    },
    "persona": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "llm-groq-qwen-qwen3-8-27b",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "groq-llm",
   "subject": "qwen/qwen3.8-27b",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies · params {\"reasoning_effort\": \"none\"}",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": null,
     "n": 12,
     "unit": "usd",
     "sampleState": "not_measured",
     "estimated": true,
     "derived": false,
     "reason": "Not shown: the provider’s list price is being re-checked"
    },
    "persona": {
     "value": 4.75,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 4.916667,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "llm-host-a-model-1",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "llm-host-a",
   "subject": "Model 1",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": 0.000168,
     "n": 12,
     "unit": "usd",
     "sampleState": "measured",
     "estimated": true,
     "derived": false,
     "reason": "list price x tokens (prices as of 2026-10-12, may be stale)"
    },
    "persona": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "llm-host-a-model-2",
   "runId": "voicebench-llm-2026-10-13",
   "providerId": "llm-host-a",
   "subject": "Model 2",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 replies · params {\"reasoning_effort\": \"none\"}",
   "metrics": {
    "ttftP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttftP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "turnP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "tokensPerSec": {
     "value": null,
     "n": 12,
     "unit": "tok_per_s",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "usdPerTurn": {
     "value": 0.000044,
     "n": 12,
     "unit": "usd",
     "sampleState": "measured",
     "estimated": true,
     "derived": false,
     "reason": "list price x tokens (prices as of 2026-10-12, may be stale)"
    },
    "persona": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    },
    "plan": {
     "value": 5,
     "n": 12,
     "unit": "score_1_5",
     "sampleState": "measured",
     "estimated": false,
     "derived": false,
     "reason": "LLM judge (Model 1 on Host A, temperature 0), mean of 1-5 ratings; known leniency, use for gross failures only"
    }
   }
  },
  {
   "id": "tts-vendor-1",
   "runId": "voicebench-tts-2026-10-13",
   "providerId": "tts-vendor-1",
   "subject": "voice (anonymized)",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "11/12 clips · 0 refusals",
   "metrics": {
    "ttfaP50": {
     "value": null,
     "n": 11,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttfaP95": {
     "value": null,
     "n": 11,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttfaMax": {
     "value": null,
     "n": 11,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "rtfP50": {
     "value": null,
     "n": 11,
     "unit": "rtf",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "werAt8k": {
     "value": 0.068,
     "n": 11,
     "unit": "ratio",
     "sampleState": "measured",
     "estimated": false,
     "derived": true,
     "reason": "mean word error rate after an 8 kHz round trip, transcribed by the STT 1 model (STT table); number formatting inflates it on the R4 line"
    }
   }
  },
  {
   "id": "tts-vendor-2",
   "runId": "voicebench-tts-2026-10-13",
   "providerId": "tts-vendor-2",
   "subject": "voice (anonymized)",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "12/12 clips · 0 refusals",
   "metrics": {
    "ttfaP50": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttfaP95": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "ttfaMax": {
     "value": null,
     "n": 12,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "rtfP50": {
     "value": null,
     "n": 12,
     "unit": "rtf",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "werAt8k": {
     "value": 0.053,
     "n": 12,
     "unit": "ratio",
     "sampleState": "measured",
     "estimated": false,
     "derived": true,
     "reason": "mean word error rate after an 8 kHz round trip, transcribed by the STT 1 model (STT table); number formatting inflates it on the R4 line"
    }
   }
  },
  {
   "id": "stt-vendor-1",
   "runId": "voicebench-stt-2026-10-13",
   "providerId": "stt-vendor-1",
   "subject": "model (anonymized)",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "34/34 clips",
   "metrics": {
    "sttFinalP50": {
     "value": null,
     "n": 34,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "sttFinalP95": {
     "value": null,
     "n": 34,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "werAt8k": {
     "value": 0.053353,
     "n": 34,
     "unit": "ratio",
     "sampleState": "measured",
     "estimated": false,
     "derived": true,
     "reason": "reference = TTS input text; includes number-normalisation mismatches"
    }
   }
  },
  {
   "id": "e2e-ref-agents-hard",
   "runId": "e2e-ref-agents-2026-10-12",
   "providerId": "ref-agents",
   "subject": "hard-scenario set",
   "dataStatus": "real",
   "status": "ok",
   "statusNote": "one measured reply gap per call (n=1 each); quiet calls; audio-measured",
   "metrics": {
    "replyGapP50": {
     "value": null,
     "n": 16,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    },
    "replyGapP95": {
     "value": null,
     "n": 16,
     "unit": "ms",
     "sampleState": "withheld",
     "estimated": false,
     "derived": false,
     "reason": "Timing figures are not published until the benchmark is re-run"
    }
   }
  }
 ],
 "corrections": []
}