ResearchTasksMulti-domain Knowledge and ReasoningMulti-domain Knowledge and ReasoningFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMMLU-ProQwen377.7Accuracy11Apr 21, 2026HLE (Humanity’s Last Exam) (official)GPT-5 pro42Exact Match7Feb 26, 2026