Modality: llm_chat · full deep dive — every ranked model, test result, and artifact.
Ranked by confidence-adjusted score (single/zero-sample, non-curated scores floored; curated empirical scores trusted as-is).
| # | Model | Provider | Adj. Score | Raw | Evidence |
|---|---|---|---|---|---|
| 1 | claude-sonnet-4-5-20250929 | anthropic | 0.530 | 0.530 | curated |
| 2 | gemini-2.5-flash | google_gemini | 0.044 | 0.116 | n=3 |
| 3 | gemini-2.5-pro | google_gemini | 0.017 | 0.017 | n=8006 |
No benchmark outputs recorded for this niche yet.