| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Safety Evaluation | Safety Evaluation Suite HarmBench, StrongReject, WildJailbreak, XSTest | HarmBench Score68.44 | 28 | |
| Safety | Safety Evaluation Suite (Salad-Bench, WildJailbreak, JailbreakBench, WildChat, WildGuard) | Safety Rate (S.R.)100 | 24 | |
| Harmful-input detection | Safety Evaluation Suite (AEG, AEG2, HB-A, HB-V, OAI, SAL, SST, Toxic, WGT, WJB) | AEG Score89.8 | 18 | |
| Safety | Safety Evaluation Suite | Score0.911 | 9 | |
| Safety Behavior Evaluation | Safety evaluation suite | ASR16.5 | 7 | |
| Safety Evaluation | Safety Evaluation Suite WJ, FH, SR | ASR (WJ)4.6 | 7 |