Medical Question Answering on MedQA (Accuracy, 95% CI, Avg. Accuracy)
82.72AccuracyPeer-reviewd reasoning
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Peer-reviewd reasoningModels=(1) + (3) + (4) + (5)2026.06 | 82.72 | 80.6 | 82 | |
| Peer-reviewd reasoningModels=(3) + (4) + (5)2026.06 | 82.64 | 80.52 | 81.6 | |
| Chain-of-thoughtModel=GPT-oss-20B2026.06 | 79.03 | 76.75 | 77.7 | |
| Chain-of-thoughtModel=Phi-42026.06 | 73.13 | 70.7 | 76.2 | |
| COT-based group votingModels=(1) + (2) + (3) + (5)2026.06 | 68.97 | 66.38 | 72.2 | |
| Chain-of-thoughtModel=Llama-3.1-8B2026.06 | 65.75 | 63.16 | 71.1 | |
| Chain-of-thoughtModel=Qwen2.5-7B2026.06 | 56.95 | 54.2 | 65 | |
| Chain-of-thoughtModel=Deepseek-llm-7b2026.06 | 31.03 | 28.52 | 43.7 |