Tongyi-MAI/MobileWorld

MobileWorld是一个更具挑战性的移动使用基准测试，旨在更好地反映现实世界的移动使用情况。它包含201个任务，覆盖20个应用程序，具有长视野、跨应用任务以及新任务类别，包括代理用户交互和MCP增强任务。MobileWorld的难点在于：1）长视野、跨应用任务，平均需要27.8个完成步骤，且62.2%的任务涉及跨应用工作流；2）新任务类别，包括代理用户交互任务（22.4%）和MCP增强任务（19.9%）。系统架构由主机和docker环境两部分组成，主机负责接收任务指令和用户交互，docker环境包含一个隔离的Android生态系统。

Mobile World is a substantially more challenging mobile-use benchmark designed to better reflect real-world mobile usage. It comprises 201 tasks across 20 applications, featuring long-horizon, cross-app tasks, and novel task categories including agent-user interaction and MCP-augmented tasks. The difficulty of Mobile World is twofold: 1) Long-horizon, cross-application tasks, requiring on average 27.8 completion steps, with 62.2% of tasks involving cross-application workflows; 2) Novel task categories, including agent-user interaction tasks (22.4%) and MCP-augmented tasks (19.9%). The system architecture consists of two main components: the host machine for task instructions and user interaction, and the docker environment containing an isolated Android ecosystem.