Sberbank has presented Russia's first generative model for creating full-fledged videos based on text descriptions - Kandinsky Video. Neural network generates a video sequence lasting up to eight seconds at 30 frames per second. Kandinsky Video consists of two blocks. One of them is responsible for creating the main frames, which later form the structure of the video plot, and the second one is responsible for generating interpolation frames, which ensure smooth movements in the video.