ResearchTasksFull-response Safety Guardrail ClassificationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedStreamSafe internal (test)SentGuard98.7F1 Score7Jun 2, 2026XSTest (test)WildGuard-7B94.7F1 Score7Jun 2, 2026Safe-RLHF (test)Qwen3Guard-8B93.2F1 Score7Jun 2, 2026BeaverTails (test)Qwen3Guard-8B84.3F1 Score7Jun 2, 2026