<a href=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/6yByXb_A099i4BHRHonNc.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/6yByXb_A099i4BHRHonNc.png\" alt=\"image\"></a></p>\n<p><a href=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/wPUrbrZ82FMYHsrf10D1s.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/wPUrbrZ82FMYHsrf10D1s.png\" alt=\"image\"></a></p>\n<p><a href=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/PNJYpH7LTLyRzlmHJbJRu.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/PNJYpH7LTLyRzlmHJbJRu.png\" alt=\"image\"></a></p>\n<p><a href=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/WS0XS-wSmKlgcJFWEj3qL.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/6388a7e98a5dbe2f3dc61faa/WS0XS-wSmKlgcJFWEj3qL.png\" alt=\"image\"></a></p>\n","updatedAt":"2026-05-21T05:56:10.560Z","author":{"_id":"6388a7e98a5dbe2f3dc61faa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6388a7e98a5dbe2f3dc61faa/Zj8pR7BG_wOQjsYzosb5p.jpeg","fullname":"Qi Mao","name":"HelenMao","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.49543264508247375},"editors":["HelenMao"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6388a7e98a5dbe2f3dc61faa/Zj8pR7BG_wOQjsYzosb5p.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.19484","authors":[{"_id":"6a0e81b4164dbbc68a26c5a1","name":"Haobo Hu","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a2","name":"Xiangwu Guo","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a3","name":"Zhiheng Chen","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a4","name":"Difei Gao","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a5","name":"Haotian Liu","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a6","name":"Libiao Jin","hidden":false},{"_id":"6a0e81b4164dbbc68a26c5a7","name":"Qi Mao","hidden":false}],"publishedAt":"2026-05-19T00:00:00.000Z","submittedOnDailyAt":"2026-05-21T00:00:00.000Z","title":"CutVerse: A Compositional GUI Agents Benchmark for Media Post-Production Editing","submittedOnDailyBy":{"_id":"6388a7e98a5dbe2f3dc61faa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6388a7e98a5dbe2f3dc61faa/Zj8pR7BG_wOQjsYzosb5p.jpeg","isPro":false,"fullname":"Qi Mao","user":"HelenMao","type":"user","name":"HelenMao"},"summary":"While GUI agents have made significant progress in web navigation and basic operating system tasks, their capabilities in professional creative workflows remain largely underexplored. To bridge this gap, we introduce Cutverse, a benchmark designed to systematically evaluate autonomous GUI agents in realistic media post-production environments. We curate expert demonstrations across 7 professional applications (e.g., Premiere Pro, Photoshop), covering 186 complex, long-horizon tasks grounded in authentic editing workflows, involving dense multimodal interfaces and tightly coupled interaction sequences. To support scalable evaluation, we develop a lightweight parser that transforms raw screen recordings and low-level interaction logs into structured, compositional GUI action trajectories with precise grounding. Extensive evaluations reveal that existing agents achieve only 36.0\\% task success on realistic media editing tasks, underscoring the challenges posed by complex, long-horizon media post-production workflows in our benchmark.While current models demonstrate promising spatial grounding, multimodal alignment, and coordinated action execution, they remain limited in long-horizon reliability and domain-specific planning.","upvotes":18,"discussionId":"6a0e81b4164dbbc68a26c5a8","githubRepo":"https://github.com/CUC-MIPG/CutVerse","githubRepoAddedBy":"user","ai_summary":"Current GUI agents show limited effectiveness in professional media post-production tasks despite advances in spatial grounding and multimodal alignment.","ai_keywords":[""],"githubStars":0,"organization":{"_id":"67dab498ed21a53369f5de73","name":"CUC-MIPG","fullname":"Multimedia Intelligent Processing Group in Communication University of China","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/640d704c8036cc2142299c19/B85B31gd7-0kjK_Rpvv3g.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"640d704c8036cc2142299c19","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/640d704c8036cc2142299c19/Wt9AslcVxWOSSc11epk8l.jpeg","isPro":true,"fullname":"Lan Chen","user":"Orannue","type":"user"},{"_id":"6603c21926220b0f3dc29549","avatarUrl":"/avatars/3d3e528f9ff4e589035c3d00f22f0aca.svg","isPro":false,"fullname":"Haobo Hu","user":"HHAObo","type":"user"},{"_id":"66f2432d5405e6677d430bbf","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/gxQWA3WTHGEQUL10fiIFh.png","isPro":false,"fullname":"GUO XIANGWU","user":"SamuelGuo","type":"user"},{"_id":"68cf444ea6e99175a057cb3a","avatarUrl":"/avatars/3ef84f5adf69847911231b1b41167be8.svg","isPro":false,"fullname":"haotian liu","user":"haotian0109","type":"user"},{"_id":"6713a356a77c43bfa414b606","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/MPZpxntH9n9F04BSIBswo.png","isPro":false,"fullname":"chk","user":"yihaitt","type":"user"},{"_id":"6a0ea5ee8f0d6eae482bcac9","avatarUrl":"/avatars/59d59ee370a32d5a310979f735b4c449.svg","isPro":false,"fullname":"luqi","user":"luqiluqi","type":"user"},{"_id":"6388a7e98a5dbe2f3dc61faa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6388a7e98a5dbe2f3dc61faa/Zj8pR7BG_wOQjsYzosb5p.jpeg","isPro":false,"fullname":"Qi Mao","user":"HelenMao","type":"user"},{"_id":"673d87d933ce9d46414ff67b","avatarUrl":"/avatars/040af60f982c4b62f666ce8296efba75.svg","isPro":false,"fullname":"yyl","user":"l13001","type":"user"},{"_id":"69aeb21297f5d3f722022d02","avatarUrl":"/avatars/440e7d41693cdaa5a9cc6c6a5e145b18.svg","isPro":false,"fullname":"Bamzzo","user":"Bamzzo","type":"user"},{"_id":"67151778ff4fc99ee4a6cdcc","avatarUrl":"/avatars/93e0e05f33a450d8e44759352b410b26.svg","isPro":false,"fullname":"LindiaC","user":"LindiaC","type":"user"},{"_id":"64aa2210e04e7f92245f54d2","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64aa2210e04e7f92245f54d2/OE43T22bLWBVgmqcJyUtu.png","isPro":false,"fullname":"Li","user":"kotion","type":"user"},{"_id":"69bd0cfe745b859b4b223c9c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Z1kfL_BmDs6YbBpI4yjuN.jpeg","isPro":false,"fullname":"WU Shiyu","user":"xie-chenxi486","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"67dab498ed21a53369f5de73","name":"CUC-MIPG","fullname":"Multimedia Intelligent Processing Group in Communication University of China","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/640d704c8036cc2142299c19/B85B31gd7-0kjK_Rpvv3g.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.19484.md"}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2605.19484 in a model README.md to link it from this page.

Cite arxiv.org/abs/2605.19484 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2605.19484 in a Space README.md to link it from this page.

Add this paper to a collection to link it from this page.