WMDP-Chem
Science · 2024-03-05
WMDP-Chem is the chemical-security split of the Weapons of Mass Destruction Proxy benchmark: four-option multiple-choice questions that proxy for hazardous knowledge relevant to chemical weapons development. Scores report accuracy; higher means more dual-use knowledge.
Top models (higher is better)
| Model | Score |
|---|---|
| Grok 4.20 | 90.0 |
| Grok 4.1 | 84.0 |
| Grok 4 | 83.0 |
| Grok 4 Fast | 77.5 |
| Grok Code Fast 1 | 52.7 |