{"slug": "rethinking-video-autoencoders-for-interactive-world-models", "title": "Rethinking Video Autoencoders for Interactive World Models", "summary": "Researchers submitted a paper to arXiv on 21 October 2026 introducing CADRE, an open family of causal diffusion representation autoencoders spanning multiple spatio-temporal compressions, built to replace text-to-video autoencoders in interactive video world models. The team trained the same world model on each CADRE autoencoder across manipulation and navigation domains with two environments each, and reports CADRE forms most of the latency-quality frontier, especially with a moving camera. The study found that at strong compression a diffusion decoder is sharper than perceptual regression, that temporal compression is more effective in the autoencoder than in the world model for robot manipulation, and that manipulation and navigation call for opposite compressions — temporal compression costs little with a still camera and spatial compression when it moves.", "body_md": "# Computer Science > Computer Vision and Pattern Recognition\n\n  [Submitted on 21 Oct 2026]\n\n# Title:Rethinking Video Autoencoders for Interactive World Models\n\n[View PDF](http://arxiv.org/pdf/2610.30129v1)\n\n[HTML (experimental)](https://arxiv.org/html/2610.30129v1)\n\nAbstract:Video world models predict future frames from past frames and actions. They serve as simulators in which agents train, plan or are evaluated, and that humans can control interactively. They generate in the latent space of a video autoencoder, whose compression sets their cost per frame and how long each action waits to take effect. Yet world models often reuse the autoencoders of text-to-video generators, built for offline generation with no agent waiting and, without their training data and code, not retrainable at another compression. We therefore ask how, where and how much a world model should compress. We build CADRE, an open family of causal diffusion representation autoencoders spanning multiple spatio-temporal compressions. Its members differ only in a small bottleneck on frozen features and share a diffusion decoder self-distilled to one or a few steps. We train the same world model on each autoencoder in two domains, manipulation and navigation, with two environments each. CADRE forms most of the latency-quality frontier, especially with a moving camera. Our study also leads to four findings. At strong compression, a diffusion decoder is sharper than perceptual regression. In robot manipulation, temporal compression is more effective in the autoencoder than in the world model. Manipulation and navigation call for opposite compressions: temporal compression costs little when the camera is still, and spatial compression when it moves. Surprisingly, with a still camera, coarser spatial tokens hurt motion far more than image quality. Code, models and a live demo are at [this https URL](https://kyutai.org/blog/2026-10-21-cadre)\n\n## Submission history\n\nFrom: Adrien Ramanana Rahary [\n[view email](http://arxiv.org/show-email/aca36454/2610.30129)]\n\n**[v1]** Wed, 21 Oct 2026 17:58:11 UTC (32,713 KB)\n\n### Additional Features\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer \n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers \n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps \n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations \n\n*(*[What are Smart Citations?](https://www.scite.ai/))\n# Code, Data and Media Associated with this Article\n\nalphaXiv \n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers \n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub \n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub \n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face \n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast \n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))\n# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower \n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender \n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\n# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [**Learn more about arXivLabs**](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/rethinking-video-autoencoders-for-interactive-world-models", "canonical_source": "http://arxiv.org/abs/2610.30129v1", "published_at": "2026-10-22 14:02:00+00:00", "updated_at": "2026-10-22 14:12:44.466297+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "computer-vision", "ai-research", "generative-ai"], "entities": ["arXiv", "CADRE", "Adrien Ramanana Rahary", "Kyutai"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/rethinking-video-autoencoders-for-interactive-world-models", "markdown": "https://wpnews.pro/news/rethinking-video-autoencoders-for-interactive-world-models.md", "text": "https://wpnews.pro/news/rethinking-video-autoencoders-for-interactive-world-models.txt", "jsonld": "https://wpnews.pro/news/rethinking-video-autoencoders-for-interactive-world-models.jsonld"}}