Safety Moderation on Aegis 2.0
86.4Prompt F1YuFeng-XGuard
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| YuFeng-XGuardSize=8B, Task=s-cls2026.05 | 86.4 | 80.4 | |
| NemotronGuardV2Size=8B, Task=s-cls2026.05 | 86.3 | 85.4 | |
| GPT-OSS-SafeGuardSize=20B, Task=s-cls2026.05 | 82.2 | 77.5 | |
| WildGuardSize=7B, Task=s-cls2026.05 | 81.5 | 82.7 | |
| GLiGuard bi-encSize=145M, Task=s-cls+IE2026.05 | 80.2 | 74.4 | |
| GLiGuard uni-encSize=147M, Task=s-cls+IE2026.05 | 80.2 | 75.7 | |
| ShieldGemmaSize=9B, Task=s-cls2026.05 | 79.9 | 74.2 | |
| GLiGuard OmniSize=209M, Task=s-cls+IE2026.05 | 79.8 | 74.6 | |
| Llama3GuardSize=8B, Task=s-cls2026.05 | 77.2 | 66.1 | |
| GLiClass Base v1.0Size=150M, Task=cls2026.05 | 74.1 | 64.5 | |
| Llama4GuardSize=12B, Task=s-cls2026.05 | 71.5 | 64.7 | |
| GLiNER2 Multi v1Size=209M, Task=cls+IE2026.05 | 71.4 | 75.7 | |
| Longformer-harmful-roSize=149M, Task=s-cls2026.05 | 52.3 | 49.8 | |
| DeBERTa-v3-base-PISize=184M, Task=p-inj2026.05 | 33.9 | 34.8 | |
| PromptGuard 2*Size=86M, Task=s-cls2026.05 | 33.7 | 35 |