ResearchTasksText-centric Visual Question Answering (Open-Ended)Text-centric Visual Question Answering (Open-Ended)FollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedText-centric VQA benchmarks (Clean)pretrained82.9AI2D Accuracy16Feb 26, 2026