Standard Operating Procedure execution on SOPBench bank 94-task (test)
65.96Success RateABSTRAL full (test)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ABSTRAL full (test)Backbone=GPT-4o, Topology=ensemble, 6 agents, Search cost=(same search), Evaluation protocol=identical to published baselines2026.03 | 65.96 | 7 |