ResearchTasksAudio QA Correctness AssessmentFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedMMAU and MMAR unseen question-based (test)ORCA (Gemma3-12B)0.91Spearman ρ18Feb 26, 2026