MedS-Bench - Information Extraction: leaderboard
The Information Extraction subtask of MedS-Bench (arXiv) — extracting structured facts from clinical text; accuracy (%). Published overview values are retained where detailed tables differ.
Metric: Accuracy (%). Source: henrychur.github.io. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.5 Sonnet (20240620) | 79.41 | #299 |
| 2 | GPT-4 | 76.92 | #438 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=meds-bench-information-extraction · How It Works · Data refreshed daily, snapshot 2026-10-20.