Medical Question Answering on MedQA USMLE (Pass@1)
97.88Pass@1MedXIAOHE
Evaluation results
| Method | Links | |
|---|---|---|
| MedXIAOHEMode=Thinking mode, Decoding=Greedy2026.02 | 97.88 | |
| GPT-5.2 ThinkingMode=Thinking mode, Decoding=Greedy2026.02 | 96.39 | |
| Gemini 3.0 ProDecoding=Greedy2026.02 | 95.52 | |
| Gemini 2.5 ProDecoding=Greedy2026.02 | 93.01 |