NVIDIA Introduces SANA-WM: A 2.6B-Parameter Open-Source World Model That Generates Minute-Scale 720p Video on a Single GPU

TL;DR AI
2 min readKey summary
NVIDIA released SANA-WM, an open-source 2.6B-parameter world model for minute-long 720p video generation.
Built from SANA-Video, it adds metric-scale 6-DoF camera control and supports single-GPU inference.
A distilled variant can generate a 60-second clip in 34 seconds on an RTX 5090.
The lower compute barrier could speed up embodied AI, simulation, and robotics research.
