ResearchTasksBinary safe/unsafe classificationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedASSE-Safety (test)BraveGuard-Qwen3-Guard-8B67.4Accuracy4Jun 2, 2026R-Judge (test)BraveGuard-Qwen3-Guard-8B57.8Accuracy4Jun 2, 2026