Wan-Streamer v0.2: Higher Resolution, Same Latency
Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.
A latency-preserving Wan Streamer upgrade: from 192p close-up calls to 640×368 clearer calls and scene-grounded mid-shot agents, still at 25 fps with ~200 ms model-side latency.<br><a href=\"https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/id-wskc0DbkMYnd_COXJH.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/id-wskc0DbkMYnd_COXJH.png\" alt=\"framework\"></a></p>\n","updatedAt":"2026-07-07T02:10:45.212Z","author":{"_id":"6522cf31777019ca30d85725","avatarUrl":"/avatars/a180b096e438e429d445b68fe703e43f.svg","fullname":"Lianghua Huang","name":"lhhuang","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":9,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8287991285324097},"editors":["lhhuang"],"editorAvatarUrls":["/avatars/a180b096e438e429d445b68fe703e43f.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.04443","authors":[{"_id":"6a4c5f6625849b193a833ff1","name":"Lianghua Huang","hidden":false},{"_id":"6a4c5f6625849b193a833ff2","name":"Zhi-Fan Wu","hidden":false},{"_id":"6a4c5f6625849b193a833ff3","name":"Yupeng Shi","hidden":false},{"_id":"6a4c5f6625849b193a833ff4","name":"Wei Wang","hidden":false},{"_id":"6a4c5f6625849b193a833ff5","name":"Mengyang Feng","hidden":false},{"_id":"6a4c5f6625849b193a833ff6","name":"Junjie He","hidden":false},{"_id":"6a4c5f6625849b193a833ff7","name":"Chen-Wei Xie","hidden":false},{"_id":"6a4c5f6625849b193a833ff8","name":"Yu Liu","hidden":false},{"_id":"6a4c5f6625849b193a833ff9","name":"Jingren Zhou","hidden":false},{"_id":"6a4c5f6625849b193a833ffa","name":"Ang Wang","hidden":false},{"_id":"6a4c5f6625849b193a833ffb","name":"Bang Zhang","hidden":false},{"_id":"6a4c5f6625849b193a833ffc","name":"Baole Ai","hidden":false},{"_id":"6a4c5f6625849b193a833ffd","name":"Chen Liang","hidden":false},{"_id":"6a4c5f6625849b193a833ffe","name":"Cheng Yu","hidden":false},{"_id":"6a4c5f6625849b193a833fff","name":"Chongyang Zhong","hidden":false},{"_id":"6a4c5f6625849b193a834000","name":"Jinwei Qi","hidden":false},{"_id":"6a4c5f6625849b193a834001","name":"Kai Zhu","hidden":false},{"_id":"6a4c5f6625849b193a834002","name":"Pandeng Li","hidden":false},{"_id":"6a4c5f6625849b193a834003","name":"Peng Zhang","hidden":false},{"_id":"6a4c5f6625849b193a834004","name":"Wenyuan Zhang","hidden":false},{"_id":"6a4c5f6625849b193a834005","name":"Xinhua Cheng","hidden":false},{"_id":"6a4c5f6625849b193a834006","name":"Yitong Huang","hidden":false},{"_id":"6a4c5f6625849b193a834007","name":"Yun Zheng","hidden":false},{"_id":"6a4c5f6625849b193a834008","name":"Yuxiang Bao","hidden":false},{"_id":"6a4c5f6625849b193a834009","name":"Yuzheng Wang","hidden":false},{"_id":"6a4c5f6625849b193a83400a","name":"Zoubin Bi","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/u4BuVOD7pttMTbQ8u1m87.mp4","https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/tbW9vra8ivsiYpkWrE7dJ.mp4","https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/b6UkvflioEhKz4AV5pEoo.mp4","https://cdn-uploads.huggingface.co/production/uploads/6522cf31777019ca30d85725/ynvOSvWNTVPEp9j0zP5IX.mp4"],"publishedAt":"2026-07-05T00:00:00.000Z","submittedOnDailyAt":"2026-07-07T00:00:00.000Z","title":"Wan-Streamer v0.2: Higher Resolution, Same Latency","submittedOnDailyBy":{"_id":"6522cf31777019ca30d85725","avatarUrl":"/avatars/a180b096e438e429d445b68fe703e43f.svg","isPro":false,"fullname":"Lianghua Huang","user":"lhhuang","type":"user","name":"lhhuang"},"summary":"We present Wan-Streamer v0.2, a latency-preserving upgrade of the native-streaming, end-to-end audio-visual interaction model. v0.2 keeps the v0.1 modeling formulation, but raises the interactive output stream from 192x336 to 640x368 while preserving approximately 200 ms model-side signal-to-signal latency at 25 FPS. The higher-resolution stream supports scene-grounded mid-shot agents whose posture, gaze, hands, nearby objects, and local scene layout remain legible during real-time conversation. To support the larger visual stream without adding user-visible delay, v0.2 keeps the thinker as a single-GPU low-latency path for streaming perception, the short language/state Transformer pass that builds the generation cache, and final decoding. The performer becomes a multi-GPU Ulysses-style context-parallel group for the expensive next-unit latent generation. Each performer rank writes incoming K/V into a pre-sharded local cache. The long high-resolution latent video sequence is split across ranks for denoising and gathered through Ulysses communication, while the much shorter audio latent sequence is generated without sequence sharding. In this split, the thinker's language/state computation reaches the performer only as K/V conditioning, so no separate language sequence has to be communicated inside the performer group. This concentrates additional hardware on visual generation while preserving the compact thinker-performer boundary, keeping total remote interaction latency at approximately 550 ms when a 350 ms bidirectional network budget is included.","upvotes":24,"discussionId":"6a4c5f6625849b193a83400b","projectPage":"https://wan-streamer.com/","ai_summary":"Wan-Streamer v0.2 enhances audio-visual interaction by increasing visual resolution while maintaining low latency through optimized thinker-performer architecture with multi-GPU parallel processing.","ai_keywords":["streaming perception","Transformer","generation cache","final decoding","Ulysses-style context-parallel group","pre-sharded local cache","denoising","Ulysses communication","visual generation","audio latent sequence","language/state computation","K/V conditioning"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"67bc7cd418dd753c02a82684","name":"Wan-AI","fullname":"Wan-AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67b610677ea7952def8b29c6/N6jQbbeaa_FcUY-wI1dgG.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6522cf31777019ca30d85725","avatarUrl":"/avatars/a180b096e438e429d445b68fe703e43f.svg","isPro":false,"fullname":"Lianghua Huang","user":"lhhuang","type":"user"},{"_id":"66615c855fd9d736e670e0a9","avatarUrl":"/avatars/0ff3127b513552432a7c651e21d7f283.svg","isPro":false,"fullname":"wangshuai","user":"wangsssssss","type":"user"},{"_id":"64d0528459503263d9fb2a2d","avatarUrl":"/avatars/902b44ccbc5074fcf1fa7da373b38f9f.svg","isPro":false,"fullname":"Zhang Wenyuan","user":"zParquet","type":"user"},{"_id":"6a2da6c8ca070ee12c6e396c","avatarUrl":"/avatars/0355287dcabaa67dbc7f0b10b87451f9.svg","isPro":false,"fullname":"Joe Mama","user":"JoeMama123123123","type":"user"},{"_id":"63bea9b987619d1458c5ea77","avatarUrl":"/avatars/f3ae4610f0d30088838b1b83683b473a.svg","isPro":false,"fullname":"Mengyang Feng","user":"archerfmy0831","type":"user"},{"_id":"64f0337e1dccbf71c3946204","avatarUrl":"/avatars/955d0533b4b7219773ef7b9a16c05848.svg","isPro":false,"fullname":"Wu Zhi-Fan","user":"wuzhifan","type":"user"},{"_id":"6463554dd2044cd1d7c6e0bf","avatarUrl":"/avatars/d7653623117268c545a7063fec69664b.svg","isPro":false,"fullname":"Bingzheng Wei","user":"Bingzheng","type":"user"},{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user"},{"_id":"69b90fb82bf8115bd3205fb0","avatarUrl":"/avatars/b285738d482c85cd19d4cde8be156a23.svg","isPro":false,"fullname":"David Miller","user":"david-miller-45678","type":"user"},{"_id":"63c1699e40a26dd2db32400d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63c1699e40a26dd2db32400d/3N0-Zp8igv8-52mXAdiiq.jpeg","isPro":false,"fullname":"Chroma","user":"Chroma111","type":"user"},{"_id":"6a4c7be96bbc08e8b9d9cab9","avatarUrl":"/avatars/a4779a234118cd2d4355c7003971a9af.svg","isPro":false,"fullname":"Leo Huang","user":"leo-huang-ai","type":"user"},{"_id":"61af81009f77f7b669578f95","avatarUrl":"/avatars/fb50773ac49948940eb231834ee6f2fd.svg","isPro":false,"fullname":"rotem israeli","user":"irotem98","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"67bc7cd418dd753c02a82684","name":"Wan-AI","fullname":"Wan-AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67b610677ea7952def8b29c6/N6jQbbeaa_FcUY-wI1dgG.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.04443.md","query":{}}">
Papers
arxiv:2607.04443
Wan-Streamer v0.2: Higher Resolution, Same Latency
Authors:
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
Abstract
Wan-Streamer v0.2 enhances audio-visual interaction by increasing visual resolution while maintaining low latency through optimized thinker-performer architecture with multi-GPU parallel processing.
Generated by Qwen/Qwen2.5-Coder-32B-Instruct
We present Wan-Streamer v0.2, a latency-preserving upgrade of the native-streaming, end-to-end audio-visual interaction model. v0.2 keeps the v0.1 modeling formulation, but raises the interactive output stream from 192x336 to 640x368 while preserving approximately 200 ms model-side signal-to-signal latency at 25 FPS. The higher-resolution stream supports scene-grounded mid-shot agents whose posture, gaze, hands, nearby objects, and local scene layout remain legible during real-time conversation. To support the larger visual stream without adding user-visible delay, v0.2 keeps the thinker as a single-GPU low-latency path for streaming perception, the short language/state Transformer pass that builds the generation cache, and final decoding. The performer becomes a multi-GPU Ulysses-style context-parallel group for the expensive next-unit latent generation. Each performer rank writes incoming K/V into a pre-sharded local cache. The long high-resolution latent video sequence is split across ranks for denoising and gathered through Ulysses communication, while the much shorter audio latent sequence is generated without sequence sharding. In this split, the thinker's language/state computation reaches the performer only as K/V conditioning, so no separate language sequence has to be communicated inside the performer group. This concentrates additional hardware on visual generation while preserving the compact thinker-performer boundary, keeping total remote interaction latency at approximately 550 ms when a 350 ms bidirectional network budget is included.
Community
A latency-preserving Wan Streamer upgrade: from 192p close-up calls to 640×368 clearer calls and scene-grounded mid-shot agents, still at 25 fps with ~200 ms model-side latency.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Get this paper in your agent:
hf papers read 2607.04443 curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Cite arxiv.org/abs/2607.04443 in a model README.md to link it from this page.
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2607.04443 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2607.04443 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.