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-4-6 | anthropic | 0.128 | 0.850 | provisional (n≤1) |
| 2 | gpt-4o | openai | 0.120 | 0.800 | provisional (n≤1) |
| 3 | gemini-2.5-pro | google_gemini | 0.112 | 0.750 | provisional (n≤1) |
| 4 | claude-haiku-4-5-20251001 | anthropic | 0.098 | 0.650 | provisional (n≤1) |
| 5 | gemini-2.5-flash | google_gemini | 0.090 | 0.600 | provisional (n≤1) |
| 6 | gpt-4o-mini | openai | 0.083 | 0.550 | provisional (n≤1) |
| 7 | grok-2-vision | xai | 0.083 | 0.550 | provisional (n≤1) |
No benchmark outputs recorded for this niche yet.