SANA-Streaming: Real-time Streaming Video Editing with Hybrid Diffusion Transformer

Researchers introduced SANA-Streaming, a real-time video-to-video editing framework built for streaming use.

It combines a hybrid diffusion transformer, cycle-reverse regularization, and GPU-focused optimizations to improve temporal consistency and throughput.

The system reports 1280×704 editing at 24 FPS end-to-end on a single RTX 5090, with the core model running at 58 FPS.

This suggests high-resolution interactive video editing can run in real time on a single consumer GPU, making live tools more practical.