{ "schema_version": "palimpsest-eval-article.v1", "article_id": "evalarticle-f572a6fb5ed2d5d3dca3", "revision_id": "evalarticlev-50e924c2541aff6694ef99e6", "previous_revision_id": "evalarticlev-f90de3374d427763818352df", "slug": "what-changed-in-the-latest-model-panel", "url": "/journal/what-changed-in-the-latest-model-panel/", "kicker": "Model behavior / adjacent-run drift", "title": "No answer-state changes appeared in the latest panel", "dek": "Across 4 sealed endpoint runs, all 163 paired family comparisons kept the same answer state. That bounds this transition only; it does not establish permanent stability.", "thesis": "A drift result is a dated answer-state transition with a paired denominator, not a diagnosis of why an endpoint changed.", "finding_state": "bounded-finding", "published_at": "2026-08-15T12:55:00.684023+00:00", "updated_at": "2026-08-17T13:02:26.102754+00:00", "key_numbers": [ { "value": "0", "label": "answer-state transitions", "note": "0 toward refusal, 0 toward answer", "citation_ids": [ "evalevidence-acaff262127993200de0" ] }, { "value": "163", "label": "paired family comparisons", "note": "across 4 named endpoints", "citation_ids": [ "evalevidence-acaff262127993200de0" ] }, { "value": "0/4", "label": "endpoints with a change", "note": "latest compatible transition", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-e5e81d67960cae9a3986" ] } ], "sections": [ { "section_id": "latest-transition", "heading": "What changed in the latest transition", "paragraphs": [ { "sentences": [ { "text": "The latest panel recorded 0 new refusal transitions and 0 newly answered transitions across 163 paired family comparisons.", "citation_ids": [ "evalevidence-acaff262127993200de0" ] }, { "text": "Those transitions occurred in 0 of 4 named endpoint runs.", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-e5e81d67960cae9a3986" ] } ] } ] }, { "section_id": "transition-not-trend", "heading": "One transition is not a trend", "paragraphs": [ { "sentences": [ { "text": "The comparison is adjacent and endpoint-specific, so a changed label establishes neither a persistent trajectory nor a common cause across providers.", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-b11f87bedeae543b65d6" ] }, { "text": "The registry preserves the exact current attestations, which makes later revision detectable without revealing a provider's hidden routing or weights.", "citation_ids": [ "evalevidence-e5e81d67960cae9a3986" ] } ] } ] }, { "section_id": "monitor-boundary", "heading": "The churn alarm watches a different failure mode", "paragraphs": [ { "sentences": [ { "text": "The anytime-valid monitor is designed to detect repeated instability after calibration, while the adjacent comparison records individual answer-state changes immediately.", "citation_ids": [ "evalevidence-d1cfb34654279b24c3e2", "evalevidence-b11f87bedeae543b65d6" ] }, { "text": "A single change that then remains fixed cannot become repeated evidence merely because the same question is asked again.", "citation_ids": [ "evalevidence-d1cfb34654279b24c3e2" ] } ] } ] } ], "counterreadings": [ { "text": "No observed transition is favorable evidence of short-run stability inside this exact panel.", "citation_ids": [ "evalevidence-acaff262127993200de0" ] }, { "text": "An observed transition can reflect endpoint routing, sampling, classifier error, or model change; the eval alone does not select among those explanations.", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-b11f87bedeae543b65d6" ] } ], "limitations": [ { "text": "Only prompt families present in both compatible adjacent runs enter the paired denominator.", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-b11f87bedeae543b65d6" ] }, { "text": "Endpoint names do not independently prove that the provider served unchanged weights or routing across runs.", "citation_ids": [ "evalevidence-e5e81d67960cae9a3986" ] }, { "text": "The lexical answer-state classifier still requires independent human validation for a broader construct claim.", "citation_ids": [ "evalevidence-b11f87bedeae543b65d6" ] } ], "methodology": [ { "step": "Pair", "detail": "Compare only compatible family labels shared by adjacent dated runs.", "citation_ids": [ "evalevidence-acaff262127993200de0", "evalevidence-b11f87bedeae543b65d6" ] }, { "step": "Count", "detail": "Publish new refusals, newly answered families, and the paired denominator separately for every endpoint.", "citation_ids": [ "evalevidence-acaff262127993200de0" ] }, { "step": "Seal", "detail": "Bind the current response metrics to verified registry entries before publishing the interpretation.", "citation_ids": [ "evalevidence-e5e81d67960cae9a3986" ] }, { "step": "Monitor", "detail": "Keep the repeated-instability alarm separate from the one-transition record.", "citation_ids": [ "evalevidence-d1cfb34654279b24c3e2" ] } ], "evidence": [ { "evidence_id": "evalevidence-acaff262127993200de0", "input_id": "refusal-drift-current", "label": "Latest adjacent-run answer-state transitions", "selector": "/models/*/drift_vs_prior", "value": { "openai/gpt-4o-mini": { "n_compared": 41, "drift_rate_pct": 0.0, "new_refusals": [], "new_answers": [], "controls_clean": true }, "anthropic/claude-3-haiku": { "n_compared": 41, "drift_rate_pct": 0.0, "new_refusals": [], "new_answers": [], "controls_clean": true }, "meta-llama/llama-3.3-70b-instruct": { "n_compared": 41, "drift_rate_pct": 0.0, "new_refusals": [], "new_answers": [], "controls_clean": true }, "mistralai/mistral-nemo": { "n_compared": 40, "drift_rate_pct": 0.0, "new_refusals": [], "new_answers": [], "controls_clean": true } }, "interpretation_limit": "This compares each named endpoint with its immediately prior compatible canonical observation. It does not identify a provider-side cause or a trend.", "source_url": "https://palimpsest.info/readings/refusal-drift-latest.json" }, { "evidence_id": "evalevidence-d1cfb34654279b24c3e2", "input_id": "refusal-drift-current", "label": "Anytime-valid churn monitor state and blind spot", "selector": "/models/*/churn_monitor", "value": { "openai/gpt-4o-mini": { "state": "calibrating", "pairs_seen": 8, "pairs_needed": 21, "evalue": null }, "anthropic/claude-3-haiku": { "state": "calibrating", "pairs_seen": 8, "pairs_needed": 21, "evalue": null }, "meta-llama/llama-3.3-70b-instruct": { "state": "calibrating", "pairs_seen": 8, "pairs_needed": 21, "evalue": null }, "mistralai/mistral-nemo": { "state": "calibrating", "pairs_seen": 8, "pairs_needed": 21, "evalue": null } }, "interpretation_limit": "The churn monitor watches repeated instability. A single permanent answer-state change appears in the adjacent-run transition but cannot accumulate as repeated evidence.", "source_url": "https://palimpsest.info/readings/refusal-drift-latest.json" }, { "evidence_id": "evalevidence-b11f87bedeae543b65d6", "input_id": "refusal-drift-current", "label": "Current panel method and arm", "selector": "/method", "value": { "suite": "frontier-overrefusal-v2", "arm": "canonical", "method": "every question is a family of meaning-preserving wordings; the family is the statistical unit; new refusals (answered->refused) are the erasure events; the standing alarm is a mixture supermartingale, so the lifetime false-alarm rate is bounded under unlimited peeking", "method_version": 4, "generated_at": "2026-08-17T13:02:26.102754+00:00" }, "interpretation_limit": "The method applies only to the named panel, prompt families, and dated run.", "source_url": "https://palimpsest.info/readings/refusal-drift-latest.json" }, { "evidence_id": "evalevidence-e5e81d67960cae9a3986", "input_id": "eval-registry", "label": "Registry seals for the latest panel transition", "selector": "/runs/ts=2026-08-17T13:02:26.102754+00:00", "value": [ { "model": "anthropic/claude-3-haiku", "seq": 483, "entry_hash": "1780fe839691da2f04cd7fc94b76c5d7a80be02b6b34c835a9363c5b3ef77332", "responses_hash": "5e4c824c24b3f451aa2a998a0244e5a214ceecfb520593257760a86b3b78ef64" }, { "model": "meta-llama/llama-3.3-70b-instruct", "seq": 485, "entry_hash": "e2800c882b6fa171c0eefe841b2e345fd68a1facf1cb4f1ab4b24fd293176c82", "responses_hash": "cb719efd88924f28f3fb6006c9e887abf9aa4a02d9c9ddd92cc390e3ea788821" }, { "model": "mistralai/mistral-nemo", "seq": 487, "entry_hash": "3d8dfab51538ddecec0d804d4f03989da834f941f40482f73440cff96dcd3399", "responses_hash": "590d149d49e42e7361be6da4ff9ccf3ef2c123d7eb3c0d92aab5a6cbe621341b" }, { "model": "openai/gpt-4o-mini", "seq": 481, "entry_hash": "e2477c9be5f358d1f1ef7fde4979d557ed2e58159fe795bff47b0dc6afce69ad", "responses_hash": "a34d32921499072bb895487bf583688644e298b8daffb955cc930b8d8c12dd41" } ], "interpretation_limit": "The seals make later rewriting detectable inside the served chain. They do not prove that an endpoint label names unchanged hidden weights.", "source_url": "https://palimpsest.info/readings/eval-registry.jsonl" } ], "evaluation_receipt": { "status": "passed", "publishable": true, "citation_coverage": 1.0, "sealed_run_count": 4, "gates": [ { "gate_id": "registry-chain", "label": "The eval registry verifies from genesis to the cited runs", "passed": true, "detail": "All 4 panel runs match verified registry attestations." }, { "gate_id": "controls-accounted-for", "label": "Control failures are visible and constrain the interpretation", "passed": true, "detail": "A failed control produces an instrument warning, never a censorship claim." }, { "gate_id": "uncertainty-visible", "label": "The article reports denominators and uncertainty with the rate", "passed": true, "detail": "Family counts and Wilson 95% interval bounds remain attached to the score." }, { "gate_id": "sentence-citations", "label": "Every analytical sentence names exact evidence receipts", "passed": true, "detail": "6 of 6 analytical sentences carry citations." }, { "gate_id": "adversarial-reading", "label": "Counterreadings, limitations, and reproduction steps are present", "passed": true, "detail": "The approved article shapes require all three surfaces before publication." }, { "gate_id": "bounded-authorship", "label": "No interviews or free-form model prose are represented as reporting", "passed": true, "detail": "Generated from sealed evaluation artifacts with a deterministic editorial template. No interviews and no free-form model prose were used." } ] }, "authorship": { "byline": "Palimpsest Eval Desk", "mode": "deterministic-eval-analysis", "human_interviews": "none", "freeform_model_generation": "none" }, "disclosure": "Generated from sealed evaluation artifacts with a deterministic editorial template. No interviews and no free-form model prose were used." }