| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| S-Eval, ORFuzzSet, and NQ Aggregated | LLM-VA | F1 Score86.81 | 72 | 5mo ago | |
| AgentHarm | TRIAD + Tri-Guard | HS Score82.79 | 15 | 1mo ago | |
| Aggregate (Koala, JBench-B, GSM-8k, SQL-1k, OrBench-H, SorryBench, JBench-H, HEX-PHI) | Average Score77.03 | 6 | 4mo ago |