ResearchDatasetsHarm CategoriesFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsSafety EvaluationHarm Categories evaluated by Claude-Opus-4-7 (Average)95.3Safety Score6Pairwise RankingSeven Harm Categories83.1Insult Pairwise Score3