Risk Identification on PaSBench
89.06RIRGPT-4o-mini
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4o-miniModel Category=Proprietary Models2026.03 | 89.06 | 44.53 | |
| HomeGuard-4BModel Category=Ours, Parameters=4B2026.03 | 86.09 | 48.43 | |
| Gemini-3-ProModel Category=Proprietary Models2026.03 | 84.38 | 58.59 | |
| HomeGuard-8BModel Category=Ours, Parameters=8B2026.03 | 83.75 | 53.91 | |
| Gemini-2.5-flashModel Category=Proprietary Models2026.03 | 79.68 | 51.56 | |
| RoboBrain2.5-8BModel Category=Open-Sourced VLMs2026.03 | 74.21 | 35.94 | |
| Qwen3-VL-235B-A22B-ThinkingModel Category=Open-Sourced VLMs2026.03 | 72.66 | 45.31 | |
| Qwen3-VL-4B-ThinkingModel Category=Open-Sourced VLMs2026.03 | 70.31 | 36.72 | |
| AgentSpecModel Category=Runtime Safeguard2026.03 | 70.31 | 44.53 | |
| Qwen3-VL-8B-ThinkingModel Category=Open-Sourced VLMs2026.03 | 69.53 | 37.5 | |
| ThinkSafeModel Category=Runtime Safeguard2026.03 | 60.94 | 39.84 | |
| PoexModel Category=Runtime Safeguard2026.03 | 52.34 | 35.94 |