MedHarm: leaderboard
MedHarm (2026) is a medical safety benchmark of 1,100 medically framed harmful requests in ten high-risk categories, such as covert poisoning, lethal dosing or weaponized pathogens, which a model should decline or redirect; a grading model scores each response. The board is the unsafe response rate, the percentage of responses containing any unsafe medical content; lower is better.
Metric: Unsafe response rate (%): share of queries whose response contains any unsafe medical content, even partial or hedged; MedHarm: 1,100 high-risk medical queries in ten categories of 110 (covert poisoning, lethal dosing, lethal drug interactions, euthanasia, illegal organ harvesting and live anesthesia, drug abuse, medicalized chemical and biological weapons, date-rape drugs, fetal harm and more), all requests a model should decline or redirect; greedy decoding, 512 new tokens; GPT-5-mini judge (91% agreement with human labels on unsafe responses); lower is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.3 Instant | 0.2 | #130 |
| 2 | Llama 3.1 8B Instruct | 1.6 | #4246 |
| 3 | GPT-5.5 | 4.1 | #49 |
| 4 | DeepSeek V4 Pro | 7.8 | #188 |
| 5 | Grok 4.3 | 8.7 | #333 |
| 6 | Qwen 2.5 7B Instruct | 20.7 | #2564 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=medharm · How It Works · Data refreshed daily, snapshot 2026-11-13.