MODUS unifies any-to-any multimodal generation with one decoder, two experts, and zero task heads. </p>\n<p><video src=\"https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/p3PRmZpRqFi7JT0dXaCMd.mp4\" controls=\"\" class=\"max-w-full!\"></video></p>\n\n<p><video src=\"https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/d7utmbJcv5Uld6xViiCPz.mp4\" controls=\"\" class=\"max-w-full!\"></video></p>","updatedAt":"2026-07-29T08:14:09.659Z","author":{"_id":"648b942c086aad6cd6c1d3a4","avatarUrl":"/avatars/b09260109859fc6909ef6bd38ecd7ff0.svg","fullname":"Mingqiao Ye","name":"mqye","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":4,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.6538339257240295},"editors":["mqye"],"editorAvatarUrls":["/avatars/b09260109859fc6909ef6bd38ecd7ff0.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.25948","authors":[{"_id":"6a69b4a29d3a1231d492b9ae","name":"Mingqiao Ye","hidden":false},{"_id":"6a69b4a29d3a1231d492b9af","name":"Zhaochong An","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b0","name":"Zhitong Gao","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b1","name":"Xian Liu","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b2","name":"François Fleuret","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b3","name":"Chuan Li","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b4","name":"Amir Zadeh","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b5","name":"Serge Belongie","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b6","name":"Afshin Dehghan","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b7","name":"Jesse Allardice","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b8","name":"David Mizrahi","hidden":false},{"_id":"6a69b4a29d3a1231d492b9b9","name":"Oğuzhan Fatih Kar","hidden":false},{"_id":"6a69b4a29d3a1231d492b9ba","name":"Roman Bachmann","hidden":false},{"_id":"6a69b4a29d3a1231d492b9bb","name":"Amir Zamir","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/xfVnochs_G6gCNx7PdzHO.png","https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/8PbZDPVqG46jmSFwmXgwz.png","https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/IG2QJZy-yyZMWoTeffg0g.png","https://cdn-uploads.huggingface.co/production/uploads/648b942c086aad6cd6c1d3a4/8ZwE4Z70cEAUQ1NfqXtmC.png"],"publishedAt":"2026-07-28T00:00:00.000Z","submittedOnDailyAt":"2026-07-29T00:00:00.000Z","title":"MODUS: Decoder-Only Any-to-Any Modeling of Diverse Modalities","submittedOnDailyBy":{"_id":"648b942c086aad6cd6c1d3a4","avatarUrl":"/avatars/b09260109859fc6909ef6bd38ecd7ff0.svg","isPro":true,"fullname":"Mingqiao Ye","user":"mqye","type":"user","name":"mqye"},"summary":"Any-to-any models predict any modality from any combination of others within a single network, a formulation used in multimodal vision and vision-language models, and increasingly in scientific domains such as ecology and astronomy. Existing any-to-any models are typically trained from scratch using encoder-decoder or diffusion architectures, impacting their performance and preventing them from using strong pre-trained decoder-only models as a prior. In this work, we investigate decoder-only any-to-any multimodal modeling, which treats all modalities symmetrically and supports arbitrary modalities as inputs and outputs without modality-specific heads, losses, or task pipelines. Because every modality is both an input and an output of the same model, the resulting model, named Modus, can support a range of applications, such as chained generation through intermediate modalities or cross-modal self-verification by scoring the model's own outputs with another generated modality. Modus demonstrates strong out-of-the-box performance and is competitive with specialist and multitask baselines using a single model across various benchmarks. All materials are open-sourced at https://modus-multimodal.epfl.ch/.","upvotes":2,"discussionId":"6a69b4a39d3a1231d492b9bc","projectPage":"https://modus-multimodal.epfl.ch/","githubRepo":"https://github.com/EPFL-VILAB/Modus","githubRepoAddedBy":"user","githubStars":16,"organization":{"_id":"624dc4110ce29222ad66abab","name":"EPFL-VILAB","fullname":"EPFL VILAB","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/WoM_Px84XpJiAU3K51Vxq.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"648b942c086aad6cd6c1d3a4","avatarUrl":"/avatars/b09260109859fc6909ef6bd38ecd7ff0.svg","isPro":true,"fullname":"Mingqiao Ye","user":"mqye","type":"user"},{"_id":"66daffbfb86f0d569a46eead","avatarUrl":"/avatars/ec1d398281eacce1b0109273315e7143.svg","isPro":true,"fullname":"Zhitong Gao","user":"ZhitongGao","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"624dc4110ce29222ad66abab","name":"EPFL-VILAB","fullname":"EPFL VILAB","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/WoM_Px84XpJiAU3K51Vxq.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.25948.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2607.25948 in a model README.md to link it from this page.

Cite arxiv.org/abs/2607.25948 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2607.25948 in a Space README.md to link it from this page.

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