ResearchTasksSkill-based Task ExecutionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSkillsBenchMUSE-Autoskill68.4Accuracy6May 27, 2026Skills-Bench robustness subset 24-task v0.3a3AIP-compiled skills0.668Mean Task Reward2Jun 4, 2026Skills-Bench 27-task headline sampleAIP-compiled skills0.705Mean Task Reward2Jun 4, 2026