Robot execution is iterative: each action changes what must be perceived, reasoned about, grounded, and verified next. This paper presents Capek 0.5, an embodied VLM that organizes post-training around four execution-facing capabilities and consolidates their specialists into a single inference-time model.</p>\n","updatedAt":"2026-08-10T08:52:31.322Z","author":{"_id":"63999a6fe657365725d0d0a4","avatarUrl":"/avatars/99736de1bc0d5decf4a6eda86e3c7937.svg","fullname":"Derek Zhe Hu","name":"zhehuderek","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9516579508781433},"editors":["zhehuderek"],"editorAvatarUrls":["/avatars/99736de1bc0d5decf4a6eda86e3c7937.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.06756","authors":[{"_id":"6a793f798e9301703eaa5e7f","name":"Ying Chen","hidden":false},{"_id":"6a793f798e9301703eaa5e80","name":"Weizhen Li","hidden":false},{"_id":"6a793f798e9301703eaa5e81","name":"Zhe Hu","hidden":false},{"_id":"6a793f798e9301703eaa5e82","name":"Zhenjiang Li","hidden":false},{"_id":"6a793f798e9301703eaa5e83","name":"Rui Jiang","hidden":false},{"_id":"6a793f798e9301703eaa5e84","name":"Zhifeng Gu","hidden":false},{"_id":"6a793f798e9301703eaa5e85","name":"Lihuang Fang","hidden":false},{"_id":"6a793f798e9301703eaa5e86","name":"Jiangping Liu","hidden":false},{"_id":"6a793f798e9301703eaa5e87","name":"Lei Yi","hidden":false},{"_id":"6a793f798e9301703eaa5e88","name":"Jie Chen","hidden":false}],"publishedAt":"2026-08-07T00:00:00.000Z","submittedOnDailyAt":"2026-08-10T00:00:00.000Z","title":"Capek 0.5: An Execution-Centric Vision-Language Model for Embodied Intelligence","submittedOnDailyBy":{"_id":"63999a6fe657365725d0d0a4","avatarUrl":"/avatars/99736de1bc0d5decf4a6eda86e3c7937.svg","isPro":false,"fullname":"Derek Zhe Hu","user":"zhehuderek","type":"user","name":"zhehuderek"},"summary":"Vision-language models are increasingly serving as the reasoning core of embodied agents. Robot execution is inherently iterative: each action reshapes the scene and physical state, continually renewing what must be perceived, reasoned about, and verified. Meeting these demands requires complementary capabilities that differ in supervision signals, prediction formats, and verification criteria. Existing approaches typically develop these capabilities against isolated, task-specific objectives, leaving open how they should be organized and integrated around execution as a whole. We present Capek 0.5, an embodied vision-language model built around an execution-centric capability taxonomy. Rather than organizing training by datasets or tasks, the taxonomy groups embodied capabilities according to their functional roles throughout execution and comprises four capability families: Spatial Reasoning, Temporal Understanding, Action Guidance, and State Verification. Each capability is first acquired by a dedicated specialist through reinforcement learning with verifiable rewards from a shared backbone, and the specialists are then consolidated into a single inference-time model through weight-space merging followed by routed policy-space distillation. We instantiate Capek 0.5 at the 2B and 35B-A3B scales and evaluate it from three complementary perspectives: comprehensive benchmark suites including Capek-StateBench, a new benchmark for state verification; a controlled study of capability retention from specialists to the unified model; and closed-loop evaluation in simulated embodied environments. Capek 0.5 improves the large majority of matched benchmark rows over its initialization, retains all four specialized capabilities in one checkpoint with quantified losses, and transfers to closed-loop embodied task execution.","upvotes":0,"discussionId":"6a793f7a8e9301703eaa5e89","projectPage":"https://xpeng-robotics.github.io/capek-0.5/","organization":{"_id":"69c9fb20951b511f5c41dd0d","name":"xpeng-robotics","fullname":"xpeng-robotics","avatar":"https://www.gravatar.com/avatar/b3437743db998befe15fce93ec62355f?d=retro&size=100"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"organization":{"_id":"69c9fb20951b511f5c41dd0d","name":"xpeng-robotics","fullname":"xpeng-robotics","avatar":"https://www.gravatar.com/avatar/b3437743db998befe15fce93ec62355f?d=retro&size=100"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.06756.md","query":{}}">
Capek 0.5: An Execution-Centric Vision-Language Model for Embodied Intelligence
Abstract
Vision-language models are increasingly serving as the reasoning core of embodied agents. Robot execution is inherently iterative: each action reshapes the scene and physical state, continually renewing what must be perceived, reasoned about, and verified. Meeting these demands requires complementary capabilities that differ in supervision signals, prediction formats, and verification criteria. Existing approaches typically develop these capabilities against isolated, task-specific objectives, leaving open how they should be organized and integrated around execution as a whole. We present Capek 0.5, an embodied vision-language model built around an execution-centric capability taxonomy. Rather than organizing training by datasets or tasks, the taxonomy groups embodied capabilities according to their functional roles throughout execution and comprises four capability families: Spatial Reasoning, Temporal Understanding, Action Guidance, and State Verification. Each capability is first acquired by a dedicated specialist through reinforcement learning with verifiable rewards from a shared backbone, and the specialists are then consolidated into a single inference-time model through weight-space merging followed by routed policy-space distillation. We instantiate Capek 0.5 at the 2B and 35B-A3B scales and evaluate it from three complementary perspectives: comprehensive benchmark suites including Capek-StateBench, a new benchmark for state verification; a controlled study of capability retention from specialists to the unified model; and closed-loop evaluation in simulated embodied environments. Capek 0.5 improves the large majority of matched benchmark rows over its initialization, retains all four specialized capabilities in one checkpoint with quantified losses, and transfers to closed-loop embodied task execution.
Community
Robot execution is iterative: each action changes what must be perceived, reasoned about, grounded, and verified next. This paper presents Capek 0.5, an embodied VLM that organizes post-training around four execution-facing capabilities and consolidates their specialists into a single inference-time model.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.06756 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.06756 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.06756 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.