Qwen Releases LiveAnimate: 14B Real-Time Streaming Human Animation Model

QwenBusinessUnit · hf · 2026-08-14

Alibaba's Qwen team introduced LiveAnimate, a model aimed at enabling real-time, long-form pose-driven human animation.

Built upon a 14B-parameter video diffusion transformer architecture, the model incorporates specialized training strategies, bounded attention caching, and sequence parallelism to improve inference efficiency and support long-form video streaming.

Original post →

More from Multimodal

Multimodal channel →