ResearchTasksMaximum reasoningMaximum reasoningFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedSynthetic Reasoning Max L=100 (test)DNC+CUW70.7Accuracy4Feb 26, 2026Synthetic Reasoning Max (L=50) (test)DNC+CUW82.3Accuracy4Feb 26, 2026