Multi-agent Planning and Composition on TaskBench Multimedia APIs (test)
91.12PrecisionAutoMAS
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| AutoMASStrategy=ReAct Style, Critique=true, Iterative Correction=true2026.05 | 91.12 | 78.88 | 84.55 | 45.24 | 0.1051 | 90.12 | 87.52 | 74.01 |