MedHELM: leaderboard
Clinically grounded medical evaluation suite built on HELM, covering medical reasoning, safety, fairness, robustness, and specialty-specific tasks.
Metric: Mean Win Rate (imported). Source: crfm.stanford.edu. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 | 66.25 | #251 |
| 2 | O3 Mini (2025-01-31) | 64.11 | #240 |
| 3 | Claude 3.7 Sonnet (20250219) | 63.57 | #201 |
| 4 | Claude 3.5 Sonnet (20241022) | 63.39 | #291 |
| 5 | GPT-4o (2024-05-13) | 56.96 | #325 |
| 6 | Gemini 2.0 Flash | 41.96 | #342 |
| 7 | GPT-4o Mini (2024-07-18) | 39.29 | #558 |
| 8 | Llama 3.3 70B Instruct | 30.36 | #534 |
| 9 | Gemini 1.5 Pro (001) | 24.11 | #444 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=medhelm · How It Works · Data refreshed daily, snapshot 2026-10-20.