Kandinsky 6.0 is here, allowing users to craft videos with audio in Full HD effortlessly! This open-source suite features two models: the advanced Pro version (29B) and the more lightweight Lite version (3B), both capable of generating videos that include synchronized dialogue, music, and ambient sounds—all without requiring any specialized skills.
Key highlights of Kandinsky 6.0 include: • The ability to produce videos lasting up to 5 seconds with accompanying audio • Lip-sync functionality, enabling characters' lips to move in harmony with spoken words • High-quality sound output at 44 kHz, adhering to streaming standards • Realistic movement physics for a more engaging experience • Available resolutions include SD, HD, and Full HD
In terms of performance: • The Pro model significantly outshines Kandinsky 5.0 in all aspects • It surpasses Veo 3.1 Fast in terms of image animation capabilities • It also outperforms LTX 2.5 regarding visual quality and speech clarity • The Lite version is compatible with consumer-grade GPUs, specifically the RTX 5060 Ti to 5090, provided they have at least 16 GB of VRAM
For developers, Kandinsky 6.0 is open-source and released under the MIT license. It can be accessed via platforms like Diffusers, FastVideo, ComfyUI, SGLang, and vLLM-omni. Anton Frolov from Sber notes, "Anyone can build their own services on top of it—completely free and without limitations."
Informational material. 18+.
