ResearchTasksModel Security EvaluationModel Security EvaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updated3,312 safety test cases (derived from 81 baseline prompts) (overall)Claude Sonnet 437.5Refusal Rate3Feb 26, 2026
3,312 safety test cases (derived from 81 baseline prompts) (overall)Claude Sonnet 437.5Refusal Rate3Feb 26, 2026