Sana System Requirements (CPU, RAM, Disk)

SANA: Efficient high-resolution image synthesis with a linear diffusion transformer. NVIDIA's open-source framework generates 4K images up to 100x faster than existing diffusion models, with a compact 0.6B-1.6B parameter architecture. Features full SD-XL training and inference pipelines, multimodal text encoders, and support for both English and Chinese prompts. Includes video generation capabilities, a web demo app, and Docker-based deployment.