Content Safety Detection on WildGuardMix
95.9F1 Scoreprompt–response verification framework
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| prompt–response verification framework2026.06 | 95.9 | 3.6 | |
| WildGuard / ShieldGemmaType=Strongest Applicable Baseline2026.06 | 90.2 | — |