HELM MedHELM - MIMIC-BHC: leaderboard
HELM MedHELM task summarizing MIMIC brief hospital courses, scored by an LLM jury.
Metric: Jury Score. Source: crfm.stanford.edu. Saturation forecast: Around January 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 4.13 | #98 |
| 2 | Claude 3.5 Sonnet (20241022) | 4.08 | #291 |
| 3 | GPT-5 Mini (2025-08-07) | 4.07 | #172 |
| 4 | O4 Mini (2025-04-16) | 4.05 | #179 |
| 5 | Claude 3.7 Sonnet (20250219) | 3.99 | #201 |
| 6 | GPT-4o (2024-05-13) | 3.97 | #325 |
| 7 | DeepSeek R1 | 3.92 | #251 |
| 8 | O3 Mini (2025-01-31) | 3.85 | #240 |
| 9 | Gemini 2.0 Flash | 3.83 | #342 |
| 10 | GPT-4o Mini (2024-07-18) | 3.79 | #558 |
| 11 | Llama 3.3 70B Instruct | 3.77 | #534 |
| 12 | Gemini 2.5 Pro (Preview 05-06) | 3.76 | #147 |
| 13 | Gemini 1.5 Pro (001) | 3.62 | #444 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-medhelm-mimic-bhc · How It Works · Data refreshed daily, snapshot 2026-10-20.