{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "An agent looks good in demos, but you cannot tell which product tasks it fails.",
    "change": "Use the study's evaluation idea to inspect a small set of tasks with human-checked answers.",
    "baseline": "Your current evaluator and a manually reviewed reference set.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Agreement with human review, missed failures, false alarms, and evaluation cost per task.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2610.20667",
  "title": "Evaluating Model Capabilities After Benchmark Saturation",
  "summary": "The method replaces discarding saturated benchmarks and retraining or building new ones, instead extracting capability signal from existing solution transcripts via pairwise judging, absolute scoring, or feature-based prediction. It costs an LLM-as-judge pipeline (API calls or a locally served judge model) plus transcript collection and validation against held-out unsaturated items, adding evaluation complexity but no model-training expense. Predictive accuracy can degrade under distribution shift: the protocols underestimated the first reasoning models released in early 2025, so judgments from older saturated items may not generalize to models with qualitatively new behaviors.",
  "rationale": "Pairwise judging runs on a single GPU or via third-party APIs and improves capability estimates from benchmarks a small team already owns, but the demonstrated failure under model-generation shifts means it must be validated on the reader's specific models before informing production decisions.",
  "technique": "Transcript-based capability evaluation on saturated benchmarks",
  "family": "avaliacao_de_agente",
  "recommendation": "testar",
  "published": "2026-10-14",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.20667",
  "independent_implementations": 1,
  "total_implementations": 2,
  "stars": 126,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "The method replaces discarding saturated benchmarks and retraining or building new ones, instead extracting capability signal from existing solution transcripts via pairwise judging, absolute scoring, or feature-based prediction. It costs an LLM-as-judge pipeline (API calls or a locally served judge model) plus transcript collection and validation against held-out unsaturated items, adding evaluation complexity but no model-training expense. Predictive accuracy can degrade under distribution shift: the protocols underestimated the first reasoning models released in early 2025, so judgments from older saturated items may not generalize to models with qualitatively new behaviors.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.20667",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Pairwise judging of saturated-benchmark transcripts reduces capability prediction error by 5.3 times relative to using saturated benchmark scores alone.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.20667",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}