ResearchTasksComprehensive ReasoningComprehensive ReasoningFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMM-StarCoVT62Accuracy18May 5, 2026ScienceQAUnsilencing Latent Reasoning84.2Accuracy18May 5, 2026Mathematical and General Reasoning Suite CombinedWIST46.7Overall Accuracy16Mar 25, 2026Aggregate BenchmarksQwen2.5-VL-7B + DUPL66.9Average Accuracy13Jun 16, 2026