Prompt Classification on HarmBench Text Prompt
98.85F1 ScoreGPT-OSS-SafeGuard-20B
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-OSS-SafeGuard-20B2025.12 | 98.85 | |
| LlamaGuard3-8B2025.12 | 98.73 | |
| GPT4o-mini2025.12 | 98.35 | |
| LlamaGuard4-12B2025.12 | 97.44 | |
| GuardReasonerVL-7B2025.12 | 96.64 | |
| GuardReasoner-8B2025.12 | 95.42 | |
| LlamaGuard3-11B-Vision2025.12 | 95.01 | |
| LlamaGuard2-8B2025.12 | 92.62 | |
| WildGuard-7B2025.12 | 92.04 | |
| ProGuard-7B2025.12 | 91.6 | |
| ProGuard-3B2025.12 | 89.2 | |
| Gemini2.5-Flash2025.12 | 87.64 | |
| LlamaGuard-7B2025.12 | 69.28 | |
| ShieldGemma-9B2025.12 | 64.18 |