Safety Classification on AegisSafety V1 (test)
92F1 ScoreQwen3Guard
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3GuardSize=8B, Mode=strict, Reasoning=false2025.12 | 92 | |
| IBM Granite Guardian 3.2Size=5B, Reasoning=false2025.12 | 88 | |
| IBM Granite Guardian 3.1Size=2B, Reasoning=false2025.12 | 87 | |
| IBM Granite Guardian 3.2Size=3B, Reasoning=false2025.12 | 87 | |
| IBM Granite Guardian 3.3Size=8B, Reasoning=false2025.12 | 87 | |
| AprielGuardSize=8B, Reasoning=true2025.12 | 86 | |
| IBM Granite Guardian 3.3Size=8B, Reasoning=true2025.12 | 85 | |
| gpt-oss-safeguardSize=20B, Reasoning=true2025.12 | 85 | |
| AprielGuardSize=8B, Reasoning=false2025.12 | 84 | |
| Qwen3GuardSize=8B, Mode=loose, Reasoning=false2025.12 | 76 | |
| ShieldGemmaSize=9B, Reasoning=false2025.12 | 74 | |
| Llama Guard 3Reasoning=false2025.12 | 70 | |
| Llama Guard 4Reasoning=false2025.12 | 64 | |
| Llama Guard 2Reasoning=false2025.12 | 63 |