ResearchTasksNon-Agentic Performance EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedFortress (test)Llama Maverick 478.75Mean Score4Mar 6, 2026HarmBench (test)LLama 4 Maverick73.92Mean Score4Mar 6, 2026Persuade (test)Gemini 2.5 Pro53.2Mean Score4Mar 6, 2026