ResearchTasksVision-Language Multi-task EvaluationVision-Language Multi-task EvaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMS COCO Unified Multi-task (test)HONES36.5VQA Score24Apr 21, 2026Qwen2.5-VL Evaluation Suite MMB, MME, POPE, SQA, VQAText (test)Vanilla82.8MMB Score10May 28, 2026