| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| HarmBench (test) | TrailBlazer | ASR (%)98.74 | 8 | 4d ago | |
| AdvBench GPT-oss-20B target (test) | TrailBlazer | ASR85.3 | 6 | 4d ago | |
| AdvBench Qwen 3-14B target (test) | PAIR | ASR51.92 | 6 | 4d ago | |
| AdvBench Gemma 3-12B target (test) | FlipAttack | ASR100 | 6 | 4d ago | |
| AdvBench Llama 3.2-11B target (test) | TrailBlazer | ASR95.51 | 6 | 4d ago |