HELM MedHELM - SHC-Conf: leaderboard
Metric: EM. Source: crfm.stanford.edu. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.7 Sonnet (20250219) | 81.1 |
| 2 | Llama 3.3 70B Instruct | 78.9 |
| 3 | GPT-4o (2024-05-13) | 77.2 |
| 4 | Claude 3.5 Sonnet (20241022) | 77 |
| 5 | Gemini 2.0 Flash | 75 |
| 6 | Gemini 1.5 Pro (001) | 74 |
| 7 | DeepSeek R1 | 73.9 |
| 8 | GPT-4o Mini (2024-07-18) | 73.8 |
| 9 | GPT-5 | 71.3 |
| 10 | O3 Mini (2025-01-31) | 71.3 |
| 11 | O4 Mini (2025-04-16) | 70.1 |
| 12 | Gemini 2.5 Pro (Preview 05-06) | 67.7 |
| 13 | GPT-5 Mini (2025-08-07) | 60.2 |
Interactive version: theaggregate.ai/benchmark?slug=helm-medhelm-shc-conf · How It Works · Data refreshed daily, snapshot 2026-09-05.