ResearchTasksSecurity EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSecEvalGPT-592.3Accuracy33Jul 3, 2026Audit AttributionYuvion VL-32B82.8Score12Jun 25, 2026Security AuditYuvion VL-32B97.4Security Audit Score12Jun 25, 2026InjecAgentNo Defense86.08Direct Harm4Apr 21, 2026