ResearchTasksRole ConsistencyFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSWE dev full set (test)ChatDev (Joint FT)8.4Total Overstepping Rate (<INFO>)4Apr 6, 2026SWE Dev hard (test)ChatDev (Joint FT)6.8Overstepping Rate (<INFO>)4Apr 6, 2026SWE easy subset dev (test)ChatDev (Joint FT)10Overstepping Rate (<INFO>)4Apr 6, 2026