ResearchTasksGUI agent successFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedWebVoyager (test)GPT-5 (SoM)90.6Success Rate18May 15, 2026WebVoyager, Online-Mind2Web, DeepShop (test average)Gemini computer-use-preview69.3Average Success Rate17May 15, 2026DeepShop (test)Orchard-GUI-4B (SFT + RL)64Success Rate17May 15, 2026
WebVoyager, Online-Mind2Web, DeepShop (test average)Gemini computer-use-preview69.3Average Success Rate17May 15, 2026