ResearchTasksSteering ValidationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSynthetic Stubs Comparison Phase 1DACS96.7Haiku Judge Accuracy4Apr 10, 2026Real Agents LLMAgent, Claude Haiku 4.5 N-scaling Phase 4 (val)DACS83.7Haiku Judge Accuracy4Apr 10, 2026
Real Agents LLMAgent, Claude Haiku 4.5 N-scaling Phase 4 (val)DACS83.7Haiku Judge Accuracy4Apr 10, 2026