ResearchDatasetsCapArenaFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsImage CaptioningCapArena-Auto91.7Average Win-Rate Margin29Open-ended GenerationCapArena11.83Average Score7Image CaptioningCapArena41.67Claude Score6Image CaptioningCapArena (test)19.15GPT-Score5