Kandinsky 6.0 Video Adds Native Audio

Native audio is the meaningful upgrade here—not another silent video endpoint. Kandinsky’s open release and Lite/Pro split make it practical for both rapid prototyping and higher-fidelity production, though five-second outputs and a 29B Pro model still impose real infrastructure costs.

• –Lite targets fast, lower-cost iteration, while Pro prioritizes cinematic quality and speech performance.
• –Synchronized audio and lip-sync reduce the need for separate sound-generation and post-production pipelines.
• –Built-in Full HD upscaling, plus standalone VSR tools, supports a cleaner path from draft generation to delivery.
• –MIT-licensed weights, code, and Diffusers integration make self-hosting and fine-tuning viable beyond fal. [Kandinsky Lab](https://www.kandinskylab.ai/)
• –Developers should benchmark latency, cost, and temporal consistency against leading closed video models before committing to production.