HELM MedHELM - MedQA: leaderboard
Metric: EM. Source: crfm.stanford.edu. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 96.82 |
| 2 | GPT-5 Mini (2025-08-07) | 95.63 |
| 3 | O4 Mini (2025-04-16) | 94.83 |
| 4 | Gemini 2.5 Pro (Preview 05-06) | 93.44 |
| 5 | O3 Mini (2025-01-31) | 92.05 |
| 6 | GPT-4o (2024-05-13) | 87.67 |
| 7 | Claude 3.5 Sonnet (20241022) | 86.48 |
| 8 | DeepSeek R1 | 85.69 |
| 9 | Claude 3.7 Sonnet (20250219) | 85.69 |
| 10 | Gemini 2.0 Flash | 84.89 |
| 11 | Llama 3.3 70B Instruct | 80.12 |
| 12 | Gemini 1.5 Pro (001) | 76.94 |
| 13 | GPT-4o Mini (2024-07-18) | 74.95 |
Interactive version: theaggregate.ai/benchmark?slug=helm-medhelm-medqa · How It Works · Data refreshed daily, snapshot 2026-09-05.