| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Safety Classification | CoSApien (test) | F1 Score94.6 | 10 | |
| Safety Alignment | CoSApien Public Prosecutor | Safety0.941 | 6 | |
| Safety Alignment | CoSApien | Safety Score99.5 | 6 | |
| Safety Alignment | CoSApien Game Development | Safety0.975 | 6 | |
| Safety Alignment | CoSApien Film Production | Safety Score0.866 | 6 | |
| Safety Alignment | CoSApien Arab Publisher | Safety91.6 | 6 | |
| Runtime Controllability | CoSApien (test) | Allow: G+/S100 | 5 | |
| Fine-grained Safety Control | CoSApien Disallowed instructions | Response Accuracy (GD)97.9 | 4 | |
| Fine-grained Safety Control | CoSApien Allowed instructions | Accuracy (GD)88 | 4 |