# Kyutai — CADRE: Rethinking Video Autoencoders for Interactive World Models

- Company: Kyutai (kyutai.org)
- Announced: 2026-10-22
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://kyutai.org/blog/2026-10-22-cadre
- Record: https://forck.live/items/26403-cadre-rethinking-video-autoencoders-for-interactive-world-models
- Subject: Moshi / Pocket TTS
- Open weights: yes
- Models affected: CADRE, DINOv3, Wan 2.2, Cosmos, LTX-2, DC-AE, MIRA, Matrix-Game 2.0, Hunyuan-GameCraft, SANA-WM, HunyuanVideo

Kyutai built CADRE, a family of open video autoencoders for interactive world models, and used them to study how a world model should compress video. The lab says world models usually borrow autoencoders from text-to-video generators, which compress time heavily and add lag that never mattered for offline clip generation. A typical 8× temporal compression delays every action by at least 400 ms at 20 frames per second, while cognitive neuroscience finds people notice their own movement is late beyond about 230 ms. CADRE reads each frame with a frozen DINOv3 image model, and a small trained bottleneck sets the compression, so family members differ only in how much they compress. Kyutai trained seven autoencoders and the same world model on each across manipulation and navigation, using 23 public video datasets totalling about 76k hours. It compared them with five baselines, including the autoencoders of Wan 2.2, Cosmos and LTX-2, and judged rollouts on image quality, motion quality and action following, with latency and frame rate timed on one H100. Kyutai released the paper, code, models and a live demo where users drive the world models from their own keyboard.

## Evidence

Verbatim from https://kyutai.org/blog/2026-10-22-cadre:

> A typical 8× temporal compression delays every action by at least 400 ms at 20 frames per second. Cognitive neuroscience finds that people notice their own movement is late beyond about 230 ms ( Shimada et al., 2009 ).

---

Record: https://forck.live/items/26403-cadre-rethinking-video-autoencoders-for-interactive-world-models
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md