HELM MedHELM - SHC-Privacy: leaderboard
Metric: EM. Source: crfm.stanford.edu. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.7 Sonnet (20250219) | 99.67 |
| 2 | GPT-4o (2024-05-13) | 98.33 |
| 3 | DeepSeek R1 | 98 |
| 4 | Llama 3.3 70B Instruct | 97.67 |
| 5 | GPT-4o Mini (2024-07-18) | 96 |
| 6 | Gemini 1.5 Pro (001) | 95.33 |
| 7 | O4 Mini (2025-04-16) | 92.33 |
| 8 | GPT-5 | 92 |
| 9 | O3 Mini (2025-01-31) | 91.67 |
| 10 | Gemini 2.0 Flash | 90.67 |
| 11 | Claude 3.5 Sonnet (20241022) | 82.67 |
| 12 | Gemini 2.5 Pro (Preview 05-06) | 82.33 |
| 13 | GPT-5 Mini (2025-08-07) | 75.67 |
Interactive version: theaggregate.ai/benchmark?slug=helm-medhelm-shc-privacy · How It Works · Data refreshed daily, snapshot 2026-09-05.