ResearchTasksGUI Agent Task SuccessFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedAndroidWorld (online)GPT-548.7Task Success Rate25May 27, 2026OSWorldUI-TARS-7B-DPO + ROSA224.4Success Rate16Mar 4, 2026OSWorld VLC (test)GPT-4o70.7Success Rate4Jun 4, 2026OSWorld Chrome (test)GPT-4o70.6Success Rate4Jun 4, 2026