Hi, I'm the author of this paper, can you add the opensource code link? <a href=\"https://github.com/InternRobotics/InternVLA-A-series\" rel=\"nofollow\">https://github.com/InternRobotics/InternVLA-A-series</a></p>\n","updatedAt":"2026-07-07T06:11:44.802Z","author":{"_id":"64a51923978242bb77ded920","avatarUrl":"/avatars/60cd68089bd39279c01f13046abb82ad.svg","fullname":"Haoxiang Ma","name":"hxma","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":4,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7702694535255432},"editors":["hxma"],"editorAvatarUrls":["/avatars/60cd68089bd39279c01f13046abb82ad.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.04988","authors":[{"_id":"6a4c712b25849b193a8340ef","user":{"_id":"64a51923978242bb77ded920","avatarUrl":"/avatars/60cd68089bd39279c01f13046abb82ad.svg","isPro":false,"fullname":"Haoxiang Ma","user":"hxma","type":"user","name":"hxma"},"name":"Haoxiang Ma","status":"claimed_verified","statusLastChangedAt":"2026-07-07T12:11:41.007Z","hidden":false},{"_id":"6a4c712b25849b193a8340f0","name":"Junhao Cai","hidden":false},{"_id":"6a4c712b25849b193a8340f1","name":"Xiaoxu Xu","hidden":false},{"_id":"6a4c712b25849b193a8340f2","name":"Hao Li","hidden":false},{"_id":"6a4c712b25849b193a8340f3","name":"Yuyin Yang","hidden":false},{"_id":"6a4c712b25849b193a8340f4","name":"Yang Tian","hidden":false},{"_id":"6a4c712b25849b193a8340f5","name":"Jiafei Cao","hidden":false},{"_id":"6a4c712b25849b193a8340f6","name":"Hongrui Zhu","hidden":false},{"_id":"6a4c712b25849b193a8340f7","name":"Zherui Qiu","hidden":false},{"_id":"6a4c712b25849b193a8340f8","name":"Zhaxizhuoma","hidden":false},{"_id":"6a4c712b25849b193a8340f9","name":"Yuqiang Yang","hidden":false},{"_id":"6a4c712b25849b193a8340fa","name":"Jiaqi Peng","hidden":false},{"_id":"6a4c712b25849b193a8340fb","name":"Xueyuan Wei","hidden":false},{"_id":"6a4c712b25849b193a8340fc","name":"Yangkun Zhu","hidden":false},{"_id":"6a4c712b25849b193a8340fd","name":"Jiahao Jiang","hidden":false},{"_id":"6a4c712b25849b193a8340fe","name":"Xing Gao","hidden":false},{"_id":"6a4c712b25849b193a8340ff","name":"Hanqing Wang","hidden":false},{"_id":"6a4c712b25849b193a834100","name":"Feng Yuan","hidden":false},{"_id":"6a4c712b25849b193a834101","name":"Kailin Li","hidden":false},{"_id":"6a4c712b25849b193a834102","name":"Xueyue Zhu","hidden":false},{"_id":"6a4c712b25849b193a834103","name":"Tai Wang","hidden":false},{"_id":"6a4c712b25849b193a834104","name":"Yan Ding","hidden":false},{"_id":"6a4c712b25849b193a834105","name":"Jiangmiao Pang","hidden":false},{"_id":"6a4c712b25849b193a834106","name":"Jia Zeng","hidden":false},{"_id":"6a4c712b25849b193a834107","name":"Jingjing Zhang","hidden":false},{"_id":"6a4c712b25849b193a834108","name":"Bowen Zhou","hidden":false},{"_id":"6a4c712b25849b193a834109","name":"Yao Mu","hidden":false},{"_id":"6a4c712b25849b193a83410a","name":"Chunhua Shen","hidden":false},{"_id":"6a4c712b25849b193a83410b","name":"Weinan Zhang","hidden":false}],"publishedAt":"2026-07-06T00:00:00.000Z","submittedOnDailyAt":"2026-07-07T00:00:00.000Z","title":"InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"Unified models for robot manipulation aim to equip one policy with both the semantic priors of pretrained VLMs and the physical dynamics learned through future prediction. In practice, existing designs tend to erode the semantics of the pretrained backbone, suffer interference among heterogeneous objectives, and learn future prediction from scratch in pixel space, leaving the dynamics priors of pretrained video generators unexploited. We present InternVLA-A1.5, which builds the policy on a native VLM backbone that keeps training on VQA and subtask prediction, and attaches a lightweight unified expert for continuous action generation. Future prediction is recast as a latent-querying problem, where a small set of learnable foresight tokens condenses the task-relevant future into a compact latent code under the supervision of a frozen pretrained video generation model, so the policy inherits world-model dynamics priors without ever learning pixel-level generation. The video branch is discarded at inference, keeping real-time control. Pretrained on 1.2M robot episodes and 3M multimodal samples, InternVLA-A1.5 achieves the best overall results on all six simulation benchmarks. In the real world, the preserved semantics deliver the strongest compositional generalization on held-out instruction bindings, and the two designs together sustain long-horizon execution.","upvotes":18,"discussionId":"6a4c712b25849b193a83410c","projectPage":"https://internrobotics.github.io/internvla-a15.github.io/","ai_summary":"InternVLA-A1.5 integrates pretrained vision-language models with future prediction in latent space to enable efficient robot manipulation with preserved semantics and long-horizon execution.","ai_keywords":["VLM backbone","future prediction","latent-querying problem","foresight tokens","pretrained video generation model","world-model dynamics priors","continuous action generation","unified expert","multimodal samples","robot episodes"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct"},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64a51923978242bb77ded920","avatarUrl":"/avatars/60cd68089bd39279c01f13046abb82ad.svg","isPro":false,"fullname":"Haoxiang Ma","user":"hxma","type":"user"},{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"649f9f0e1dcdd7e9e07edee6","avatarUrl":"/avatars/64fc858b40c3822ebf471e652144e433.svg","isPro":false,"fullname":"yuanfengjeff","user":"yuanfengjeff","type":"user"},{"_id":"67e695e6a56a7e73b23336b1","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67e695e6a56a7e73b23336b1/o2rlNClx2AQF3aikxVqg1.png","isPro":false,"fullname":"Xiaoxu Xu","user":"xiaoxuxxxx","type":"user"},{"_id":"675b91e7a86e54985542f9ba","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/675b91e7a86e54985542f9ba/JVt4lmWplJTj8khQPThre.jpeg","isPro":false,"fullname":"Modi Shi (SII)","user":"ModiShi","type":"user"},{"_id":"682955ea6b63a6eb59f21fae","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/682955ea6b63a6eb59f21fae/nwTr3cl4jL4OuW5P6Fwge.jpeg","isPro":false,"fullname":"Jiahao Jiang","user":"zaleni","type":"user"},{"_id":"64905cd589f22918ecaca080","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/2S7I7uZL49CXbUN2T7p63.jpeg","isPro":false,"fullname":"Xiefan Guo","user":"xiefan-guo","type":"user"},{"_id":"66eea12392d33025023cba7d","avatarUrl":"/avatars/e7adfc10f44f045ee833126093517d07.svg","isPro":false,"fullname":"paul_dicks","user":"menwhatcanisay","type":"user"},{"_id":"65c4eb7cd1dcbd30d86febec","avatarUrl":"/avatars/001c8f02e8ce794b2c21883628b2da72.svg","isPro":false,"fullname":"free-bit","user":"free-bit","type":"user"},{"_id":"647f3aece9c81260ff86c843","avatarUrl":"/avatars/04489491bf435e19bfdcbdfd80204eb1.svg","isPro":false,"fullname":"Chao Zhou","user":"ASHIDAKA","type":"user"},{"_id":"699d6f620d9ed77547c2ba64","avatarUrl":"/avatars/2b3538042ce5c14f1c17e13e1d20253e.svg","isPro":false,"fullname":"Domend886","user":"Domend886","type":"user"},{"_id":"664d930f4b870dd167473c1c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/664d930f4b870dd167473c1c/TXVEPGvkhftdI_xE1mluu.jpeg","isPro":false,"fullname":"Andy Guan","user":"andytonglove","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.04988.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2607.04988 in a dataset README.md to link it from this page.

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