Medical Question Answering on HealthBench Hard 1000
86AccuracyClaude Sonnet 4.6
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Sonnet 4.6Type=Commercial LLMs, Setting=BioMedArena2026.05 | 86 | |
| Gemini 3.1 ProType=Commercial LLMs, Setting=BioMedArena2026.05 | 80.8 | |
| Gemini 3 FlashType=Commercial LLMs, Setting=BioMedArena2026.05 | 80.7 | |
| Claude Opus 4.6Type=Commercial LLMs, Setting=BioMedArena2026.05 | 80.2 | |
| Claude Opus 4.5Type=Commercial LLMs, Setting=BioMedArena2026.05 | 78.9 | |
| Claude Opus 4.6Type=Commercial LLMs, Setting=Baseline2026.05 | 76.2 | |
| Claude Sonnet 4.5Type=Commercial LLMs, Setting=BioMedArena2026.05 | 75.3 | |
| Claude Opus 4.5Type=Commercial LLMs, Setting=Baseline2026.05 | 72.5 | |
| Gemini 3.1 ProType=Commercial LLMs, Setting=Baseline2026.05 | 72 | |
| Claude Sonnet 4.6Type=Commercial LLMs, Setting=Baseline2026.05 | 69.3 | |
| Claude Sonnet 4.5Type=Commercial LLMs, Setting=Baseline2026.05 | 67.9 | |
| Gemini 3 FlashType=Commercial LLMs, Setting=Baseline2026.05 | 59.7 |