SANA-Video 2.0 launched, promising efficient high-quality video generation
Enze Xie announced the model on X, highlighting a hybrid attention backbone and a new training pipeline that cuts compute while preserving quality.
By Ryan Merket · Published · Updated
Primary source: X
Why it matters
Efficient video generation lowers compute costs and expands the range of applications that can incorporate high‑fidelity AI video, reshaping the market for developers and enterprises.
On July 24th 2026, machine‑learning practitioner Enze Xie (@xieenze_jr) unveiled SANA-Video 2.0 in a five‑part X thread. The post positions the system as a "full‑stack optimized video model designed for efficiency — while still delivering high quality" and outlines the technical choices that differentiate it from existing generators.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.