Structured Distillation of Web Agent Capabilities Enables Generalization
Open paper
Benchmark
Model
Rank
Results
computer-use-agents-on-visualwebarena
A3-Qwen3.5-9B
#2
Success Rate: 33.9
computer-use-agents-on-webarena
A3-Qwen3.5-9B
#4
Success Rate: 41.5
Rank counts only results with a code link.