LongCat Avatar is built on the LongCat-Video foundation - a 13.6 billion parameter video generation model developed by Meituan's LongCat research team.

The model unifies Text-to-Video, Image-to-Video, and Video-Continuation tasks within a single framework, enabling minutes-long video generation without quality degradation.

- No cold starts - instant API access - 13.6B parameter model for exceptional quality - Unified architecture for consistent performance - REST API for easy integration