Prompt Classification on Aegis Text Prompt 2.0
83.52F1 ScoreGPT4o-mini
Evaluation results
| Method | Links | |
|---|---|---|
| GPT4o-miniModel Type=Closed-source2025.12 | 83.52 | |
| ProGuard-7BModel Type=VLM-based guard2025.12 | 83.38 | |
| ProGuard-3BModel Type=VLM-based guard2025.12 | 82.51 | |
| GuardReasonerVL-7BModel Type=VLM-based guard2025.12 | 82.4 | |
| Gemini2.5-FlashModel Type=Closed-source2025.12 | 82.01 | |
| GuardReasoner-8BModel Type=LLM-based guard2025.12 | 81.56 | |
| GPT-OSS-SafeGuard-20BModel Type=LLM-based guard2025.12 | 78.6 | |
| LlamaGuard2-8BModel Type=LLM-based guard2025.12 | 77.76 | |
| LlamaGuard3-8BModel Type=LLM-based guard2025.12 | 77.06 | |
| LlamaGuard-7BModel Type=LLM-based guard2025.12 | 76.06 | |
| LlamaGuard3-11B-VisionModel Type=VLM-based guard2025.12 | 74.72 | |
| LlamaGuard4-12BModel Type=VLM-based guard2025.12 | 72.95 | |
| ShieldGemma-9BModel Type=LLM-based guard2025.12 | 72.05 | |
| WildGuard-7BModel Type=LLM-based guard2025.12 | 66.58 |