FramePack

FramePack is a groundbreaking video diffusion technology that brings high-quality AI video generation to everyday consumer GPUs — you only need 6GB of VRAM to get started. By efficiently packing frame context and using constant-length input, FramePack makes next-frame prediction fast and accessible without requiring expensive cloud resources. It runs entirely locally on your own machine, maintains consistent visual quality over long video sequences, and uses a bi-directional sampling approach for smooth, coherent results. Professional-grade video generation, finally within reach for creators and developers without enterprise-level hardware.