The Feed12.10.2026github.com/kandinskylab

Open Kandinsky now ships picture and sound together

Kandinsky 6.0 Video is MIT-licensed and generates five-second clips with synced audio and lip sync in one pass.

Open Kandinsky now ships picture and sound together

If you want joint video and audio without a closed API, this is the open drop to try. Kandinsky Lab released Kandinsky 6.0 Video under MIT: a 29B Pro and a 3B Lite that render five-second clips with synchronised 44 kHz audio, including lip sync, from text or image.

Diffusers weights are on Hugging Face, and day-one packs land in ComfyUI and FastVideo. Treat Pro as a workstation job and Lite as the local test bed, then decide whether native sound-in-the-same-pass beats your current cut-and-dub stack.

The FeedNext storyAn AI short just won a $450,000 festival Grand Prix

Stay ahead of creative AI.

New editions when they drop. The best of the Feed in between. One email, no noise.