ResearchTasksMulti-Image Visual ReasoningFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedBLINKQwen3-VL69.1Accuracy51Jun 30, 2026MIMICOurs (masked)0.755Common Accuracy10Feb 26, 2026