MedS-Bench - Diagnosis: leaderboard
The Diagnosis subtask of MedS-Bench (arXiv) — inferring diagnoses from case descriptions; accuracy (%). Published overview values are retained where detailed tables differ.
Metric: Accuracy (%). Source: henrychur.github.io. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet (20240620) | 60.24 | #299 |
| 2 | GPT-4 | 58.13 | #438 |
Interactive version: theaggregate.ai/benchmark?slug=meds-bench-diagnosis · How It Works · Data refreshed daily, snapshot 2026-10-20.