The model card describes two text- or image-conditioned video-and-audio models: Lite, with 3 billion parameters, and Pro, with 29 billion. It says both generate five-second clips with synchronized 44 kHz audio. The repository provides a Diffusers example for the Pro checkpoint and identifies a separate super-resolution pipeline that can raise output resolution to Full HD. Access to the model files requires users to sign in and accept the repository conditions. The model card documents the release and usage path; it does not establish comparative quality or independent evaluation results.
Raw SignalOct 7, 20261 min read1 source
Kandinsky Lab has published the Kandinsky 6.0 Video model family on Hugging Face
In one minute
Kandinsky Lab has published the Kandinsky 6.0 Video model family on Hugging Face.