Response Classification on Aegis Text Response 2.0
86.3F1 ScoreNemoGuard-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| NemoGuard-8BSafeguard Type=Post-hoc2026.02 | 86.3 | |
| Qwen3Guard-4B-GenSafeguard Type=Post-hoc2026.02 | 85.9 | |
| Qwen3Guard-0.6B-GenSafeguard Type=Post-hoc2026.02 | 85.2 | |
| Qwen3Guard-8B-GenSafeguard Type=Post-hoc2026.02 | 85.1 | |
| NExT-GuardSafeguard Type=Streaming2026.02 | 82.9 | |
| WildGuard-7BSafeguard Type=Post-hoc2026.02 | 82.7 | |
| Qwen3Guard-4B-StreamSafeguard Type=Streaming2026.02 | 82.7 | |
| ProGuard-7B2025.12 | 82.27 | |
| NExT-Guard*Safeguard Type=Streaming2026.02 | 82 | |
| KelpSafeguard Type=Streaming2026.02 | 81.5 | |
| Qwen3Guard-8B-StreamSafeguard Type=Streaming2026.02 | 81.1 | |
| MD-Judge-7B2025.12 | 80.57 | |
| ProGuard-3B2025.12 | 80.51 | |
| Gemini2.5-Flash2025.12 | 80.44 | |
| SCM-7BSafeguard Type=Streaming2026.02 | 80.4 | |
| DynaGuard-8B2025.12 | 80.34 | |
| Qwen3Guard-0.6B-StreamSafeguard Type=Streaming2026.02 | 80 | |
| GPT4o-mini2025.12 | 79.22 | |
| GuardReasoner-8B2025.12 | 79.13 | |
| GPT-OSS-SafeGuard-20B2025.12 | 77.66 | |
| GuardReasonerVL-7B2025.12 | 74.64 | |
| ShieldGemma-27BSafeguard Type=Post-hoc2026.02 | 73.9 | |
| LlamaGuard2-8B2025.12 | 73.53 | |
| LlamaGuard-7B2025.12 | 72.58 | |
| LlamaGuard3-8B2025.12 | 72.33 | |
| LlamaGuard4-12B2025.12 | 71.82 | |
| WildGuard-7B2025.12 | 71.45 | |
| LlamaGuard3-11B-Vision2025.12 | 69.63 | |
| ShieldGemma-9BSafeguard Type=Post-hoc2026.02 | 69.3 | |
| LlamaGuard3-8BSafeguard Type=Post-hoc2026.02 | 64.8 | |
| ShieldGemma-9B2025.12 | 64.21 | |
| LlamaGuard4-12BSafeguard Type=Post-hoc2026.02 | 62.2 |