SciHazard (Science-Specific Models) - Oversafety: leaderboard
SciHazard (Shanghai AI Laboratory, 2026) measures scientific misuse risk with 2,400 hazardous questions grounded in regulated substances and documented misuse scenarios across 12 disciplines, plus 600 benign questions that reuse the same trigger words. Over-refusal is the share of the 600 benign defence, detection and emergency-response questions that are refused, in percent; lower is better; this board covers science-specific models and agents.
Metric: Over-refusal rate (%; share of 600 benign defence, detection and emergency-response questions that reuse hazardous trigger words and are refused; lower is better; proprietary science-specific models and agents). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternAgent | 0 | |
| 2 | S1-VL-32B | 3.3 | |
| 3 | Intern-S1-Pro | 10.8 | #573 |
Interactive version: theaggregate.ai/benchmark?slug=scihazard-science-specific-models-oversafety · How It Works · Data refreshed daily, snapshot 2026-11-13.