Arena

Head-to-head on the public 60

Compare two runs on individual tasks. Pick a pair, filter by pass/fail outcomes and app, then step through both trajectories side by side - each step's screen, thought, tool calls and token spend - to see exactly where the two runs diverged.

Attribution

If you refer to that Arena design, please cite MobileWorld: 

@inproceedings{kong2025mobileworld,
  title={MobileWorld: Benchmarking Autonomous Mobile Agents in Agent-User Interactive, and MCP-Augmented Environments},
  author={Quyu Kong and Xu Zhang and Zhenyu Yang and Nolan Gao and Chen Liu and Panrong Tong and Chenglin Cai and Hanzhang Zhou and Jianan Zhang and Liangyu Chen and Zhidan Liu and Steven Hoi and Yue Wang},
  booktitle={Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL)},
  year={2026},
  url={https://arxiv.org/abs/2512.19432},
}