Kandinsky LabKandinsky 6.0 Video
Kandinsky 6.0 Video Lite
This 3B diffusion model accepts text and optional image conditioning. It generates five-second clips with 44 kHz sound, including speech synchronization.
Choose the training variant
The standard checkpoint uses flow-matching generation.kandinskylab/Kandinsky-6.0-Lite-5s-Diffusers — model card and files
Keep audio and resolution handling explicit
Full-HD output uses a separate super-resolution model. The pipeline can disable audio sampling. Prompt expansion uses its Qwen text encoder and does not establish another Kandinsky model.kandinskylab/Kandinsky-6.0-Lite-5s-Diffusers — model card and files
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | kandinskylab/Kandinsky-6.0-Lite-5s-Diffusers |
| Generator size | 3B parameterskandinskylab/Kandinsky-6.0-Lite-5s-Diffusers — model card and files |
| Checkpoint role | Standard generationkandinskylab/Kandinsky-6.0-Lite-5s-Diffusers — model card and files |
| Weight license | MITkandinskylab/Kandinsky-6.0-Lite-5s-Diffusers — model card and files |