Question Answering on MMLU Clinical Knowledge (test)
63.21AccuracyPubMed Reasoner
Evaluation results
| Method | Links | |
|---|---|---|
| PubMed ReasonerBackbone Model=GPT2026.03 | 63.21 | |
| PubMed ReasonerBackbone Model=Gemini2026.03 | 61.36 | |
| Self-reflectionBackbone Model=GPT2026.03 | 60.79 | |
| Self-reflectionBackbone Model=Gemini2026.03 | 60.67 | |
| GPT4oRetrieval/Agent Strategy=Without retrieval, Source=leaderboard2026.03 | 60.64 | |
| Gemini-2.5 ProRetrieval/Agent Strategy=Without retrieval2026.03 | 60.52 | |
| RAG methodBackbone Model=Gemini2026.03 | 58.94 | |
| RAG methodBackbone Model=GPT2026.03 | 58.26 |