ResearchBenchmarksWeb Task Automation on WebArenaFollow22.2AccuracyAgentLab4.8329.34113.8518.359Oct 28, 2025Evaluation ResultsMethodMethodLinksAccuracyAgentLabModel=Qwen-2.5-14B-Cod...Model=Qwen-2.5-14B-Coder-Instruct, Training Data=ADP Data2025.1022.2AgentLabModel=Qwen-2.5-14B-Cod...Model=Qwen-2.5-14B-Coder-Instruct, Training Data=–2025.105.5