ResearchTasksGeneral-domain capability retentionGeneral-domain capability retentionFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedAIME-2024, MATH-500, GSM8K, HumanEval, MMLUBase62.9AIME-2024 Accuracy21May 19, 2026