CausVid is an advanced video generation model that achieves instant video frame generation by adapting a pre-trained bidirectional diffusion transformer into a causal transformer. This technology is significant as it greatly reduces the latency of video generation, allowing for interactive frame rates (9.4 FPS) when streaming on a single GPU. The CausVid model supports generation from text to video as well as zero-shot image-to-video generation, showcasing a new pinnacle in video generation technology.