ResearchTasksBlack-box system prompt extractionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast Updated41 Models Closed-weightJUSTASK4.8Avg Turns2Feb 26, 202641 Models OverallJUSTASK2.8Average Turns1Feb 26, 202641 Models Open-weightJUSTASK2.4Avg Turns1Feb 26, 2026