[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-1-en-105":3,"doc-seo-381174-105":53,"doc-detail-381174-en":126},{"code":4,"msg":5,"data":6},0,"success",[7,14,19,24,29,34,39,44,49],{"id":8,"doc_module":9,"doc_module_name":10,"category_name":11,"show_sort_weight":12,"slug":13},11,1,"Template","Presentations",90,"presentations",{"id":15,"doc_module":9,"doc_module_name":10,"category_name":16,"show_sort_weight":17,"slug":18},12,"Resumes",80,"resumes",{"id":20,"doc_module":9,"doc_module_name":10,"category_name":21,"show_sort_weight":22,"slug":23},14,"Invoices",70,"invoices",{"id":25,"doc_module":9,"doc_module_name":10,"category_name":26,"show_sort_weight":27,"slug":28},15,"Posters",60,"posters",{"id":30,"doc_module":9,"doc_module_name":10,"category_name":31,"show_sort_weight":32,"slug":33},16,"Social Media",50,"social-media",{"id":35,"doc_module":9,"doc_module_name":10,"category_name":36,"show_sort_weight":37,"slug":38},17,"Forms",40,"forms",{"id":40,"doc_module":9,"doc_module_name":10,"category_name":41,"show_sort_weight":42,"slug":43},18,"Letters",30,"letters",{"id":45,"doc_module":9,"doc_module_name":10,"category_name":46,"show_sort_weight":47,"slug":48},21,"Paper Templates",5,"papers-templates",{"id":50,"doc_module":9,"doc_module_name":10,"category_name":51,"show_sort_weight":4,"slug":52},158,"General","general-158",{"code":4,"msg":54,"data":55},"ok",{"site_id":56,"language":57,"slug":58,"title":59,"keywords":60,"description":61,"schema_data":62,"social_meta":119,"head_meta":121,"extra_data":123,"updated_unix":125},105,"en","infiniteweb-scalable-web-environment-synthesis-for-gui-agent-training","INFINITEWEB - Scalable Web Environment Synthesis for GUI Agent Training","","GUI agents that operate graphical interfaces for users require suitable training environments, yet such environments are scarce. INFINITEWEB automatically generates functional web environments at scale to support GUI agent training. Unlike single-page generation, realistic multi-page websites need consistency, correctness, and diversity across interlinked states. The system uses unified specification, task-centric test-driven development, and seed plus reference design images, and produces verifiable task evaluators for dense reinforcement learning rewards.",{"@graph":63,"@context":118},[64,80,101],{"@type":65,"itemListElement":66},"BreadcrumbList",[67,71,74,77],{"item":68,"name":69,"@type":70,"position":9},"https://docshare.wps.com","Home","ListItem",{"item":72,"name":10,"@type":70,"position":73},"https://docshare.wps.com/template/",2,{"item":75,"name":11,"@type":70,"position":76},"https://docshare.wps.com/template/presentations/",3,{"item":78,"name":59,"@type":70,"position":79},"https://docshare.wps.com/template/infiniteweb-scalable-web-environment-synthesis-for-gui-agent-training/381174/",4,{"url":78,"name":59,"@type":81,"image":82,"author":87,"headline":59,"publisher":90,"fileFormat":93,"inLanguage":57,"description":61,"dateModified":94,"datePublished":95,"encodingFormat":93,"isAccessibleForFree":96,"interactionStatistic":97},"DigitalDocument",{"url":83,"@type":84,"width":85,"height":86},"https://docshare.wps.com/thumbnails/infiniteweb-scalable-web-environment-synthesis-for-gui-agent-training/381174.png","ImageObject",442,249,{"name":88,"@type":89},"Olivia Brown","Person",{"url":68,"name":91,"@type":92},"DocShare","Organization","application/pdf","2026-10-28","2026-09-24",true,{"@type":98,"interactionType":99,"userInteractionCount":8},"InteractionCounter",{"@type":100},"ViewAction",{"@type":102,"mainEntity":103},"FAQPage",[104,110,114],{"name":105,"@type":106,"acceptedAnswer":107},"What problem does INFINITEWEB address for GUI agent training?","Question",{"text":108,"@type":109},"It addresses the scarcity of suitable training environments by automatically generating functional web environments at scale for GUI agents.","Answer",{"name":111,"@type":106,"acceptedAnswer":112},"How does INFINITEWEB ensure consistency across multiple interconnected web pages?",{"text":113,"@type":109},"It uses unified specification: deriving a complete set of data models and interfaces from tasks, then generating all pages from the shared specification.",{"name":115,"@type":106,"acceptedAnswer":116},"How does INFINITEWEB improve correctness and enable reinforcement learning?",{"text":117,"@type":109},"It applies task-centric test-driven development by deriving test cases from task specifications and iteratively refining code until task-relevant tests pass, and it generates verifiable task evaluators for dense reward signals.","https://schema.org",{"og:url":78,"og:type":120,"og:title":59,"og:site_name":91,"og:description":61},"article",{"robots":122,"canonical":78},"index,follow",{"doc_id":124,"site_id":56},381174,1791729136,{"code":4,"msg":5,"data":127},{"doc_id":124,"user_id":128,"nickname":88,"user_avatar":129,"doc_module":9,"category_id":8,"category_name":11,"doc_title":59,"doc_description":61,"doc_content":130,"file_id":131,"file_url":132,"file_type":133,"file_size":134,"view_count":8,"is_deleted":4,"is_public":9,"is_downloadable":9,"audit_status":9,"page_count":135,"language":136,"language_code":57,"site_id":56,"html_lang":57,"table_of_contents":137,"faqs":138,"seo_title":139,"seo_description":61,"update_tm":140,"read_time":141},16904993612988,"https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd","INFINITEWEB: Scalable Web Environment Synthesis for GUI Agent  \nTraining  \nZiyun Zhang1* Zongyu Guo3  \nZezhou Wang2* Xiaoyi Zhang3† Jiahao Li3 Bin Li3 Yan Lu3  \n1Peking University 2Nanjing University  \n3Microsoft Research Asia  \nAbstract  \nGUI agents that interact with graphical interfaces on behalf of users represent a promising direction for practical AI assistants. However, training such agents is hindered by the scarcity of suitable environments. We present INFINITEWEB, a system that automatically generates functional web environments at scale for GUI agent training. While LLMs perform well on generating a single webpage, building a realistic and functional website with many interconnected pages faces challenges. We address these challenges through unified specification, task-centric test-driven development, and a combination of website seed with reference design image to ensure diversity. Our system also generates verifiable task evaluators enabling dense reward signals for reinforcement learning. Experiments show that INFINITEWEB surpasses commercial coding agents at realistic website construction, and GUI agents trained on our generated environments achieve significant performance improvements on OSWorld, Online-Mind2Web, and MobileWorld, demonstrating the effectiveness of the proposed system. Our code and generated websites are publicly available at [https:](https:)//[github.com/microsoft/FIVE-UI-Evol](github.com/microsoft/FIVE-UI-Evol).  \n1 Introduction  \nGUI agents, autonomous systems that interact with graphical user interfaces to complete tasks on behalf of users, have emerged as a promising direction for building practical AI assistants (Xie et al., 2024 ; Zhou et al., 2024) . Recent advances (Hong et al., 2024 ; Qin et al., 2025) have demonstrated that vision-language models can be end-toend trained with reinforcement learning algorithms as GUI agents to understand screenshots, reason about UI elements, and execute human-like actions  \n∗ : Equal contribution and work done during the internship at Microsoft Research Asia. † : Project lead.  \nFigure 1: GUI agent performance improves with more training data generated by INFINITEWEB. Dashed lines indicate potential for further scaling.  \nto automate tasks in digital world. However, training such agents remains challenging due to the scarcity of suitable environments.  \nExisting GUI agent benchmarks, such as MiniWoB++ (Liu et al., 2018), WebArena (Zhou et al., 2024), and OSWorld (Xie et al., 2024), provide valuable testbeds but suffer from fundamental limitations in scale and diversity as training environments. These benchmarks are manually constructed, requiring significant human effort to design websites or download applications, define tasks, and create evaluation criteria. As a result, they contain only tens to hundreds of applications, insufficient for training agents that can generalize across the vast diversity of real-world websites. Although recent work (Sun et al., 2025 ; Xu et al., 2024 ; Xie et al., 2025a) proposes synthesizing tasks or trajectories, these approaches still operate within the same benchmark environments, limiting model training on a small set of specific applications.  \nA natural question arises: Can we automatically generate environments for GUI agent training? While large language models (LLMs) have shown remarkable code generation capabilities (Chenet al., 2022 ; Si et al., 2025 ; Jimenez et al., 2023), especially for web frontend (Leviathan et al., 2026), directly applying them to generate complete, func-  \n28465  \nProceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages 28465–28492  \nJuly 2-7, 2026 ©2026 Association for Computational Linguistics  \nFigure 2: A showcase of diverse websites generated by INFINITEWEB, spanning various categories including e-commerce, booking platforms, social media, and more.  \ntional websites faces three critical challenges.  \nGenerating such e","cbCaid8s4MrAtsMU","https://ap.wps.com/l/cbCaid8s4MrAtsMU","pdf",13541464,28,"English","# Abstract\n# Introduction\n## Motivation and Challenges\n## Proposed Approach: INFINITEWEB\n## Unified Specification\n## Task-Centric Test-Driven Development\n## Diversity via Seed and Reference Images","[{\"question\":\"What problem does INFINITEWEB address for GUI agent training?\",\"answer\":\"It addresses the scarcity of suitable training environments by automatically generating functional web environments at scale for GUI agents.\"},{\"question\":\"How does INFINITEWEB ensure consistency across multiple interconnected web pages?\",\"answer\":\"It uses unified specification: deriving a complete set of data models and interfaces from tasks, then generating all pages from the shared specification.\"},{\"question\":\"How does INFINITEWEB improve correctness and enable reinforcement learning?\",\"answer\":\"It applies task-centric test-driven development by deriving test cases from task specifications and iteratively refining code until task-relevant tests pass, and it generates verifiable task evaluators for dense reward signals.\"}]","INFINITEWEB - Scalable Web Environment Synthesis for GUI Agent Training | PDF",1790243962,10]