Safety Detection on MAS Gemma-2-9B (Current)
95.35AccuracyAgentLens
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AgentLensMethod Category=Interpretability-based2026.06 | 95.35 | 86.67 | |
| Gemini-3-ProMethod Category=LLM-as-a-Judge2026.06 | 89.53 | 75.68 | |
| LLaMA-Guard-4-12BMethod Category=Guardrail-based2026.06 | 88.37 | 58.33 | |
| Claude Sonnet-4.6Method Category=LLM-as-a-Judge2026.06 | 87.21 | 71.79 | |
| Deepseek-V4-FlashMethod Category=LLM-as-a-Judge2026.06 | 87.21 | 71.79 | |
| Deepseek-V4-ProMethod Category=LLM-as-a-Judge2026.06 | 87.21 | 70.27 | |
| Claude Opus-4.6Method Category=LLM-as-a-Judge2026.06 | 86.05 | 70 | |
| AGrail (GPT-4o)Method Category=Guardrail-based2026.06 | 80.23 | 62.22 | |
| GPT-4oMethod Category=LLM-as-a-Judge2026.06 | 63.95 | 47.46 | |
| GPT-5.2Method Category=LLM-as-a-Judge2026.06 | 61.63 | 45.9 |