ResearchTasksAgent SafetyFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedR-JudgeGemini-3.1-Pro97.3Accuracy92Jun 2, 2026ASSEBenchDRAFT (Qwen3Guard-Gen-4B)92.04Accuracy69Jun 2, 2026AuraGenExtractor95.31Accuracy47Apr 7, 2026AgentHarmTrustAgent4PAIR ASR6Jun 5, 2026