Medical Question Answering on HealthBench normal
65.2Pass@1MedXIAOHE
Evaluation results
| Method | Links | |
|---|---|---|
| MedXIAOHEMode=Thinking mode, Decoding=Greedy2026.02 | 65.2 | |
| GPT-5.2 ThinkingMode=Thinking mode, Decoding=Greedy2026.02 | 63.34 | |
| Gemini 2.5 ProDecoding=Greedy2026.02 | 49.8 | |
| Gemini 3.0 ProDecoding=Greedy2026.02 | 48.4 |