ResearchTasksLeave-one-corpus-out transferFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast Updatedwebrun size (flat)76.6AUC3Jun 23, 2026tau2-benchrun size (flat)72.5AUC3Jun 23, 2026OpenHandsrun size (flat)0.699AUC3Jun 23, 2026SWE-Gymflat plus dependency layer (f+d)59.9AUC3Jun 23, 2026tau-benchflat plus dependency layer (f+d)0.567AUC3Jun 23, 2026