Graphical user interfaces (GUIs) are the primary medium for human-computer interaction, yet automating GUI interactions remains challenging due to the complexity of visual elements, dynamic environments, and the need for multi-step reasoning. Existing methods based on vision-language models (VLMs) often suffer from limited resolution, domain mismatch, and insufficient sequential decisionmaking capability. To address these issues, we propose Mano, a robust GUI agent built upon a multi-modal foundation model pre-trained on extensive web and computer system data. Our approach integrates a novel simulated environment for high-fidelity data generation, a three-stage training pipeline (supervised fine-tuning, offline reinforcement learning, and online reinforcement learning), and a verification module for error recovery. Mano demonstrates state-of-the-art performance on multiple GUI benchmarks, including Mind2Web and OSWorld, achieving significant improvements in success rate and operational accuracy. Our work provides new insights into the effective integration of reinforcement learning with VLMs for practical GUI agent deployment, highlighting the importance of domain-specific data, iterative training, and holistic reward design.
@article{arxiv.2509.17336,
title = {Mano Technical Report},
author = {Tianyu Fu and Anyang Su and Chenxu Zhao and Hanning Wang and Minghui Wu and Zhe Yu and Fei Hu and Mingjia Shi and Wei Dong and Jiayao Wang and Yuyang Chen and Ruiyang Yu and Siran Peng and Menglin Li and Nan Huang and Haitian Wei and Jiawei Yu and Yi Xin and Xilin Zhao and Kai Gu and Ping Jiang and Sifan Zhou and Shuo Wang},
journal= {arXiv preprint arXiv:2509.17336},
year = {2025}
}