{
  "denominator": 60,
  "runs": [
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022067,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-first-pass-findings.json",
      "experimentId": "solar-decide-native-fresh1-p0",
      "id": "solar-decide-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 55,
        "follow_up_needed": 58,
        "sentiment": 56,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "First pass only; later Solar repeats are separate. The linked report has source-backed paired changes.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "52c2ef28b7156d1846e7ccef962a8318e6c5ae6d708a1b7782ee45d4472afb41",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/first-pass.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 467.469690796
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 441340,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022655,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-first-pass-findings.json",
      "experimentId": "solar-decide-native-fresh1-p0",
      "id": "solar-decide-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 53,
        "follow_up_needed": 60,
        "sentiment": 54,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": "solar-decide-native-fresh1-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "First pass only; later Solar repeats are separate. The linked report has source-backed paired changes.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "52c2ef28b7156d1846e7ccef962a8318e6c5ae6d708a1b7782ee45d4472afb41",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/first-pass.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 485.012149418
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 453100,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.023159,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-first-pass-findings.json",
      "experimentId": "solar-decide-native-fresh1-p0",
      "id": "solar-decide-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 53,
        "follow_up_needed": 59,
        "sentiment": 56,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": "solar-decide-native-fresh1-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "First pass only; later Solar repeats are separate. The linked report has source-backed paired changes.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "52c2ef28b7156d1846e7ccef962a8318e6c5ae6d708a1b7782ee45d4472afb41",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/first-pass.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 546.665764957
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 463180,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022067,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh2-p0",
      "id": "solar-decide-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 53,
        "follow_up_needed": 58,
        "sentiment": 56,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 413.03287796
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 441340,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022655,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh2-p0",
      "id": "solar-decide-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 52,
        "follow_up_needed": 59,
        "sentiment": 53,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": "solar-decide-native-fresh2-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 422.241402585
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 453100,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.023159,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh2-p0",
      "id": "solar-decide-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 52,
        "follow_up_needed": 58,
        "sentiment": 56,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": "solar-decide-native-fresh2-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 461.055465961
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 463180,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022067,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh3-p0",
      "id": "solar-decide-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 57,
        "testimonial_potential": 57
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 469.723400829
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 441340,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.022655,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh3-p0",
      "id": "solar-decide-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 59,
        "sentiment": 54,
        "serious_concern_reported": 57,
        "testimonial_potential": 56
      },
      "model": "upstage/solar-decide",
      "pairedEligible": true,
      "parentBaselineId": "solar-decide-native-fresh3-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 487.070015999
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 453100,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": false,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.02277075,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.1048576
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/solar-decide-full-findings.json",
      "experimentId": "solar-decide-native-fresh3-p0",
      "id": "solar-decide-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 52,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "upstage/solar-decide",
      "pairedEligible": false,
      "parentBaselineId": "solar-decide-native-fresh3-p0",
      "protocolId": "solar-decide-native-choice-v1",
      "provider": "Upstage",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "The original run and its continuation returned 59 usable answers. DEV-009 has no returned answer and retains a separate unknown-cost bound. This is not a clean 60-answer repeat.",
      "returnedModel": "upstage/solar-decide-20260928",
      "savedResponses": 59,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 59,
        "totalRequests": 60,
        "totalSeconds": 473.12019091
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": false,
        "input": 455415,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 236,
        "reasoning": null,
        "reportedRequests": 59,
        "totalRequests": 60
      },
      "valid": 59
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01568912,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh1-p0",
      "id": "liquid-d1-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 43,
        "follow_up_needed": 55,
        "sentiment": 49,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "be428da40efef67a9db9bc8334f176c45f235fdbd02af8c128d93f188d61dbb9",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh1/P0/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 392228,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01618832,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh1-p0",
      "id": "liquid-d1-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 42,
        "follow_up_needed": 54,
        "sentiment": 48,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh1-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "c533a647ad9c6005f9fc823456586833f0a8e6767aaeda0ade189757a6c2b548",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh1/P1/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 404708,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01662032,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh1-p0",
      "id": "liquid-d1-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 41,
        "follow_up_needed": 54,
        "sentiment": 47,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh1-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "c6b6c96bd6716a1e832c347d34ee7792c03b15be80d15ca533dd227a1674c7c1",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh1/P2/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 415508,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01568912,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh2-p0",
      "id": "liquid-d1-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 43,
        "follow_up_needed": 55,
        "sentiment": 49,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "02439ad17b8a7cbe490657161daf919ed27ddb8c6eb8194207f9017f0bb4a871",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh2/P0/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 392228,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01618832,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh2-p0",
      "id": "liquid-d1-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 54,
        "sentiment": 50,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh2-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "8115ce76db82c8e670fa41dc4ecd28b02f0f53ee70fcaac2468870ff1ad98c51",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh2/P1/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 404708,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01662032,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh2-p0",
      "id": "liquid-d1-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 41,
        "follow_up_needed": 54,
        "sentiment": 47,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh2-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "a8658452cd4dc1d1b2fb67780e6d6ae2acfbbc59c929d77b2af0e855e2567eee",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh2/P2/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 415508,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01568912,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh3-p0",
      "id": "liquid-d1-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 43,
        "follow_up_needed": 55,
        "sentiment": 49,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "b7a7765a46fe9212e2fc697cf511328741f49ac63d1c9fa94b4e6e52c040c3b1",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh3/P0/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 392228,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01618832,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh3-p0",
      "id": "liquid-d1-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 42,
        "follow_up_needed": 55,
        "sentiment": 48,
        "serious_concern_reported": 57,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh3-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "a2053dce965b3fabe5907821768b8fee9565dde899e62c1275b9e1ca3e04cf57",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh3/P1/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 404708,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01662032,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/liquid-d1-native-full-findings.json",
      "experimentId": "liquid-d1-native-fresh3-p0",
      "id": "liquid-d1-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 41,
        "follow_up_needed": 54,
        "sentiment": 47,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "liquid/d1",
      "pairedEligible": true,
      "parentBaselineId": "liquid-d1-native-fresh3-p0",
      "protocolId": "liquid-d1-native-choice-v1",
      "provider": "Liquid",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "liquid/d1-20260930",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "cc64cafa20bd827715da10d36226f216faee46a506d744fd01072d5b85aed740",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/liquid-d1-native-v1/full-v1/fresh3/P2/development.public.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 415508,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016342872,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh1-p0",
      "id": "tev1-4b-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 51,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 389116,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016836792,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh1-p0",
      "id": "tev1-4b-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh1-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 400876,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.017267712,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh1-p0",
      "id": "tev1-4b-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh1-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 411136,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016342872,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh2-p0",
      "id": "tev1-4b-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 51,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 389116,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016836792,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh2-p0",
      "id": "tev1-4b-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh2-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 400876,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.017267712,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh2-p0",
      "id": "tev1-4b-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh2-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 411136,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016342872,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh3-p0",
      "id": "tev1-4b-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 51,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 389116,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.016836792,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh3-p0",
      "id": "tev1-4b-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh3-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 400876,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.017267712,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/public-site/tev-native-full-findings.json",
      "experimentId": "tev1-4b-native-fresh3-p0",
      "id": "tev1-4b-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 44,
        "follow_up_needed": 56,
        "sentiment": 52,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "togethercomputer/tev1-4b-experimental",
      "pairedEligible": true,
      "parentBaselineId": "tev1-4b-native-fresh3-p0",
      "protocolId": "tev1-4b-native-choice-v1",
      "provider": "Together",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed native-choice repeat. The linked report has source-backed paired and repeat comparisons.",
      "returnedModel": "togethercomputer/tev1-4b-experimental-20260923",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/tev-native-v1/full-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 411136,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 480,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03184656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh1-p0",
      "id": "clef-openrouter-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 56,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 48.012163953
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03472656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh1-p0",
      "id": "clef-openrouter-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 57,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh1-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 44.769508831
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03718896,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh1-p0",
      "id": "clef-openrouter-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh1-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 46.624966168
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 154954,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03184656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh2-p0",
      "id": "clef-openrouter-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 56,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 45.382607703
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03472656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh2-p0",
      "id": "clef-openrouter-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 57,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh2-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 47.920633706
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03718896,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh2-p0",
      "id": "clef-openrouter-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh2-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 48.902634543
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 154954,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03184656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh3-p0",
      "id": "clef-openrouter-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 56,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 49.789267415
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03472656,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh3-p0",
      "id": "clef-openrouter-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 57,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh3-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 51.735369749
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.03718896,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-openrouter-native-fresh3-p0",
      "id": "clef-openrouter-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 55,
        "testimonial_potential": 56
      },
      "model": "cloudflare/clef",
      "pairedEligible": true,
      "parentBaselineId": "clef-openrouter-native-fresh3-p0",
      "protocolId": "clef-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 47.02971608
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 154954,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01194246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh1-p0",
      "id": "clef-flash-openrouter-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 50,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 35.443486958
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01302246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh1-p0",
      "id": "clef-flash-openrouter-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 47,
        "follow_up_needed": 57,
        "sentiment": 52,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": "clef-flash-openrouter-native-fresh1-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 35.098615872
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01394586,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh1-p0",
      "id": "clef-flash-openrouter-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 46,
        "follow_up_needed": 57,
        "sentiment": 50,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": "clef-flash-openrouter-native-fresh1-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 31.39665429
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 154954,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01194246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh2-p0",
      "id": "clef-flash-openrouter-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 50,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 28.704386917
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01302246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh2-p0",
      "id": "clef-flash-openrouter-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 47,
        "follow_up_needed": 57,
        "sentiment": 52,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": "clef-flash-openrouter-native-fresh2-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 31.499787499
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01394586,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh2-p0",
      "id": "clef-flash-openrouter-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 46,
        "follow_up_needed": 57,
        "sentiment": 50,
        "serious_concern_reported": 56,
        "testimonial_potential": 58
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": "clef-flash-openrouter-native-fresh2-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 30.813134166
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 154954,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01194246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh3-p0",
      "id": "clef-flash-openrouter-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 57,
        "sentiment": 50,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 29.918418998
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 132694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01302246,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh3-p0",
      "id": "clef-flash-openrouter-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 47,
        "follow_up_needed": 57,
        "sentiment": 52,
        "serious_concern_reported": 56,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": true,
      "parentBaselineId": "clef-flash-openrouter-native-fresh3-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 30.998951418
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 144694,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": false,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01371321,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.02359296
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "clef-flash-openrouter-native-fresh3-p0",
      "id": "clef-flash-openrouter-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 45,
        "follow_up_needed": 56,
        "sentiment": 49,
        "serious_concern_reported": 55,
        "testimonial_potential": 57
      },
      "model": "cloudflare/clef-flash",
      "pairedEligible": false,
      "parentBaselineId": "clef-flash-openrouter-native-fresh3-p0",
      "protocolId": "clef-flash-openrouter-native-choice-v1",
      "provider": "Cloudflare",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Final prompt run combines 38 original answers with 21 exact never-sent continuation answers. DEV039 remains a provider failure with unknown cost; 59 of 60 positions have usable answers.",
      "returnedModel": "cloudflare/clef-flash",
      "savedResponses": 59,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 59,
        "totalRequests": 60,
        "totalSeconds": 35.07448383
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": false,
        "input": 152369,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 59,
        "totalRequests": 60
      },
      "valid": 59
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0133523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh1-p0",
      "id": "luna-decisions-openrouter-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 54,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 23.660274667
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 133523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0145523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh1-p0",
      "id": "luna-decisions-openrouter-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 54
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh1-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 22.319072122
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 145523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0155783,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh1-p0",
      "id": "luna-decisions-openrouter-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 57,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 53
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh1-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 21.074326376
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 155783,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0133523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh2-p0",
      "id": "luna-decisions-openrouter-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 54,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 27.928647704
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 133523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0145523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh2-p0",
      "id": "luna-decisions-openrouter-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 54
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh2-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 26.847528129
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 145523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0155783,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh2-p0",
      "id": "luna-decisions-openrouter-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 57,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 53
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh2-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 28.399992168
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 155783,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0133523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh3-p0",
      "id": "luna-decisions-openrouter-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 58,
        "sentiment": 54,
        "serious_concern_reported": 56,
        "testimonial_potential": 56
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 27.129368458
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 133523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0145523,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh3-p0",
      "id": "luna-decisions-openrouter-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 51,
        "follow_up_needed": 58,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 54
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh3-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 28.92583717
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 145523,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.0155783,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/findings.json",
      "experimentId": "luna-decisions-openrouter-native-fresh3-p0",
      "id": "luna-decisions-openrouter-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 49,
        "follow_up_needed": 57,
        "sentiment": 55,
        "serious_concern_reported": 59,
        "testimonial_potential": 53
      },
      "model": "openai/gpt-6-luna-decisions",
      "pairedEligible": true,
      "parentBaselineId": "luna-decisions-openrouter-native-fresh3-p0",
      "protocolId": "luna-decisions-openrouter-native-choice-v1",
      "provider": "OpenAI",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed OpenRouter native Choice run. The linked report shows its exact source evidence and available paired or repeat comparisons.",
      "returnedModel": "openai/gpt-6-luna-decisions-20261006",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/clef-openrouter-v1/findings-v1/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "Client request elapsed sum covers returned answers only. It includes transport and service time, not isolated inference time; per-request percentiles are unavailable.",
        "p95Seconds": null,
        "requests": 60,
        "totalRequests": 60,
        "totalSeconds": 30.70255054
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 155783,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 0,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01542736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh1-p0",
      "id": "perplexity-decider-native-fresh1-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 385684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01590736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh1-p0",
      "id": "perplexity-decider-native-fresh1-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 58,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh1-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 397684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01631776,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh1-p0",
      "id": "perplexity-decider-native-fresh1-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh1-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh1",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 407944,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01542736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh2-p0",
      "id": "perplexity-decider-native-fresh2-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 385684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01590736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh2-p0",
      "id": "perplexity-decider-native-fresh2-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 58,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh2-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 397684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01631776,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh2-p0",
      "id": "perplexity-decider-native-fresh2-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh2-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh2",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 407944,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P0",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01542736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh3-p0",
      "id": "perplexity-decider-native-fresh3-p0",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": null,
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 385684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P1",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01590736,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh3-p0",
      "id": "perplexity-decider-native-fresh3-p1",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 59,
        "sentiment": 58,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh3-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 397684,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    },
    {
      "complete": true,
      "condition": "P2",
      "cost": {
        "actualUsd": null,
        "estimatedUsd": null,
        "knownUsd": 0.01631776,
        "note": "Known development inference charge covers returned answers only; smoke tests excluded. Unknown-cost bounds are retained separately, and no invoice amount is asserted.",
        "unknownUpperBoundUsd": 0.0
      },
      "effort": "not applicable",
      "evidenceUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/findings.json",
      "experimentId": "perplexity-decider-native-fresh3-p0",
      "id": "perplexity-decider-native-fresh3-p2",
      "interface": "four native Choice questions",
      "metrics": {
        "all_four": 54,
        "follow_up_needed": 58,
        "sentiment": 59,
        "serious_concern_reported": 57,
        "testimonial_potential": 59
      },
      "model": "perplexity/pplx-decider-v1-27b",
      "pairedEligible": true,
      "parentBaselineId": "perplexity-decider-native-fresh3-p0",
      "protocolId": "perplexity-decider-native-choice-v1",
      "provider": "Perplexity",
      "records": 60,
      "repeatPass": "fresh3",
      "resultStatus": "Closed Perplexity native Choice run. Scores compare with provisional reference labels; linked records support prompt and repeat comparisons.",
      "returnedModel": "perplexity/pplx-decider-v1-27b-20261001",
      "savedResponses": 60,
      "sourceOnlyDetails": true,
      "sourceRecordSha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035",
      "sourceRecordsUrl": "https://github.com/adambkovacs/candidate-experience-benchmark/blob/main/results/perplexity-decider-v1/full-v2/public-projection.json",
      "surface": "OpenRouter native Choice",
      "timing": {
        "inferenceReportedRequests": 0,
        "inferenceSeconds": null,
        "kind": "record",
        "medianSeconds": null,
        "note": "No public client-duration series or server inference duration is available for this run.",
        "p95Seconds": null,
        "requests": 0,
        "totalRequests": 60,
        "totalSeconds": null
      },
      "tokens": {
        "cacheWrite": null,
        "cachedInput": null,
        "complete": true,
        "input": 407944,
        "note": "Provider-reported development usage covers the returned answers only. Missing token categories are unavailable, not zero.",
        "output": 240,
        "reasoning": null,
        "reportedRequests": 60,
        "totalRequests": 60
      },
      "valid": 60
    }
  ],
  "schema": "supplemental-decision-runs-v1",
  "scope": "Closed public native Choice development runs only. Individual predictions stay in linked public projections; no private responses or reference labels are embedded.",
  "sources": [
    {
      "path": "data/pilot/proposed_labels.jsonl",
      "sha256": "440fa16759473b6d4ff52fe7e7296e5f2dfca0a58f5df26f20aef0daafed1464"
    },
    {
      "path": "public-site/data-provider-errors-v1.json",
      "sha256": "52d827b39690291be224523b7e493657706ea746a2dea30fe88f4b62c2c2f1b7"
    },
    {
      "path": "public-site/liquid-d1-native-full-findings.json",
      "sha256": "77dbda1f74de139a1ac3bdc3b1f8b2069672704c108025b3e5e2d1ec33c895c6"
    },
    {
      "path": "public-site/solar-decide-first-pass-findings.json",
      "sha256": "f1bed1b6d8392d5017db5efaaf560b6695103a859493938395c0529258967bb9"
    },
    {
      "path": "public-site/solar-decide-full-findings.json",
      "sha256": "c7d40ec31f573983e4a0f1153add5679191f935a446844ced733cf55c69f2216"
    },
    {
      "path": "public-site/tev-native-full-findings.json",
      "sha256": "53fc9caee090a8d3cb511083ba57baba145820a5f7c65050a81e10935576e5e7"
    },
    {
      "path": "results/clef-openrouter-v1/findings-v1/findings.json",
      "sha256": "1f32791a76b8487eef634adc8a12d6d3179698e3cfcd3c8e5098a349a2ec1377"
    },
    {
      "path": "results/clef-openrouter-v1/findings-v1/public-projection.json",
      "sha256": "81bb51e9d95b92124642d7efa6a546b29873772dd72b400fc2b695409789bae6"
    },
    {
      "path": "results/clef-openrouter-v1/findings-v1/public-projection.receipt.json",
      "sha256": "063d417a9bd91a3139f890232a9b5ed3291b368aa80fb7f5aecd941147260619"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh1/P0/development.public.json",
      "sha256": "be428da40efef67a9db9bc8334f176c45f235fdbd02af8c128d93f188d61dbb9"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh1/P1/development.public.json",
      "sha256": "c533a647ad9c6005f9fc823456586833f0a8e6767aaeda0ade189757a6c2b548"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh1/P2/development.public.json",
      "sha256": "c6b6c96bd6716a1e832c347d34ee7792c03b15be80d15ca533dd227a1674c7c1"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh2/P0/development.public.json",
      "sha256": "02439ad17b8a7cbe490657161daf919ed27ddb8c6eb8194207f9017f0bb4a871"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh2/P1/development.public.json",
      "sha256": "8115ce76db82c8e670fa41dc4ecd28b02f0f53ee70fcaac2468870ff1ad98c51"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh2/P2/development.public.json",
      "sha256": "a8658452cd4dc1d1b2fb67780e6d6ae2acfbbc59c929d77b2af0e855e2567eee"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh3/P0/development.public.json",
      "sha256": "b7a7765a46fe9212e2fc697cf511328741f49ac63d1c9fa94b4e6e52c040c3b1"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh3/P1/development.public.json",
      "sha256": "a2053dce965b3fabe5907821768b8fee9565dde899e62c1275b9e1ca3e04cf57"
    },
    {
      "path": "results/liquid-d1-native-v1/full-v1/fresh3/P2/development.public.json",
      "sha256": "cc64cafa20bd827715da10d36226f216faee46a506d744fd01072d5b85aed740"
    },
    {
      "path": "results/perplexity-decider-v1/full-v2/findings.json",
      "sha256": "3b738756fb63ce4406f9eba18f759ee09708d50f57ffaf4d2d8ee05eafb67ef5"
    },
    {
      "path": "results/perplexity-decider-v1/full-v2/plan.json",
      "sha256": "fb1bbd6b49f925ad370f0746050738d01bec4528b942eaaf0c450eaee8773ea8"
    },
    {
      "path": "results/perplexity-decider-v1/full-v2/public-projection.json",
      "sha256": "0afd1a31adc6171cd013170b456be2c26b4480a8e1eb873493c5c423ebad7035"
    },
    {
      "path": "results/perplexity-decider-v1/plan.json",
      "sha256": "275d8fdca723719882331d72566c93b131b3bee40b5097ed5497902f06835815"
    },
    {
      "path": "results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.json",
      "sha256": "58e64df0e6d1d5534cdceae58b2ca826f9cd946c5195423474124837680f0426"
    },
    {
      "path": "results/solar-decide-native-full-v1/execution-adapter-v2/all-nine.public-projection.receipt.json",
      "sha256": "a6cf69402f75630eb9c48691d4a05f463e75a8930a431e49109ff574c07be78e"
    },
    {
      "path": "results/solar-decide-native-full-v1/execution-adapter-v2/first-pass.public-projection.json",
      "sha256": "52c2ef28b7156d1846e7ccef962a8318e6c5ae6d708a1b7782ee45d4472afb41"
    },
    {
      "path": "results/solar-decide-native-full-v1/execution-adapter-v2/first-pass.public-projection.receipt.json",
      "sha256": "ca439ec504181f94f3c387872961265a278ffc0071d18e19151f61032dbe7f75"
    },
    {
      "path": "results/tev-native-v1/full-v1/public-projection.json",
      "sha256": "0a90f0ef00ee699910ada1826ec2d23834e936301bb90df8e6826082fd43e732"
    },
    {
      "path": "results/tev-native-v1/full-v1/public-projection.receipt.json",
      "sha256": "c57ffff4130adf913a10f6e70b4b259e69aa5943847023291038d46bfade3a92"
    },
    {
      "path": "scripts/build_supplemental_decision_runs_v1.py",
      "sha256": "70a6d47f076fba4626533f67b113c9dd0034941acd12ad8340ff3b7159e7410b"
    }
  ]
}