Safety Risk Attribution on Risk Attribution C
98.6ScoreGPT-5.4
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5.4Model Size=≈1.5T2026.06 | 98.6 | |
| K2.5Model Size=≈1T2026.06 | 88.3 | |
| Qwen3.5-PlusModel Size=397B2026.06 | 87.2 | |
| Yuvion VL-32BProtocol=Instruct2026.06 | 85.4 | |
| Qwen3-VL-32BProtocol=Instruct2026.06 | 85 | |
| Qwen3-VL-32BProtocol=Thinking2026.06 | 83.9 | |
| Yuvion VL-8BProtocol=Instruct2026.06 | 83.6 | |
| Yuvion VL-32BProtocol=Reasoning2026.06 | 83.2 | |
| Qwen3-VL-8BProtocol=Instruct2026.06 | 81.2 | |
| Yuvion VL-8BProtocol=Reasoning2026.06 | 81 | |
| Qwen3-VL-8BProtocol=Thinking2026.06 | 80.8 | |
| GLM-5Model Size=≈700B2026.06 | 22.7 |