ResearchTasksText-based safety moderationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedBeaverTailsRoblox Guard 1.087.3F1 Score60Jun 23, 2026WildGuardSingGuard-2B80.26F1 Score26Jun 23, 2026OpenAIGPT-4o82.3F1 Score26Feb 26, 2026Toxic ChatGGuard82F1 Score24Feb 26, 2026AegisOMNIGUARD-7B84F1 Score12Feb 26, 2026