MedES - Drug Safety: leaderboard
Medication-safety set of MedES (Jin et al., Shanghai Jiao Tong University, AAAI 2026): 3,896 questions selected from the NLPEC licensed-pharmacist exam corpus on dosage limits, contraindicated populations and drug interactions. Accuracy in percent; higher is better.
Metric: Accuracy (%; 3,896 medication-safety questions selected from NLPEC). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 | 88.4 | #467 |
| 2 | DeepSeek V3 | 85.9 | #705 |
| 3 | GPT-4 Turbo | 69.2 | #1002 |
| 4 | GPT-4 | 65.9 | #996 |
| 5 | GPT-3.5 | 45.8 | #2430 |
| 6 | DeepSeek-R1-Distill-Qwen-7B | 34.7 | #8045 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=medes-drug-safety · How It Works · Data refreshed daily, snapshot 2026-11-12.