HarmBench
Red-team and refusal test
safety
Definition
- Version
- not in the notes
- Category
- safety
- Maintainer
- Academic
- Status
- not in the notes
- Home page
- not in the notes
Sources
- huggingface.co/mistralai/Shieldstral-1.0-3B https://huggingface.co/mistralai/Shieldstral-1.0-3B
Scores
| Model | Score | Who ran it | Setting | View or population | Date | Source |
|---|---|---|---|---|---|---|
 Shieldstral 1.0 3B | 99.4 F1 | vendor-reported | – | – | – | huggingface.co |
Where this comes from
Scores from different settings, harnesses or dates are not directly comparable; the row says which. Sorted by score, highest first, as a reading aid, not a ranking.
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "HarmBench", https://aurion.semai.id/benchmarks/harmbench/, records of October 2026.