MUA-RL: Multi-turn User-interacting Agent Reinforcement Learning for agentic tool use
Open paper
Benchmark
Model
Rank
Results
agents-on-bench-airline
MUA-RL-32B
–
Accuracy: 45.4
agents-on-bench-retail
MUA-RL-32B
–
Accuracy: 67.3
Rank counts only results with a code link.