Kandinsky 3.1

Kandinsky 3.1 is an AI model by Sber designed to generate images and videos from a text prompt or a sketch.

Where you can use Kandinsky

• Fusion Brain
• Telegram
• VKontakte
• ruDALL-E

What it can do

Kandinsky generates digital visuals from descriptions and can also animate uploaded images. If you specify a genre or style in the prompt, the model aims to follow those conventions.

Access and key features

Online access is available via the official website and through bots in social networks and messengers, including VKontakte and Telegram. It can also be used through the SberBank Online and Salut apps.

• Free generation from Russian-language prompts
• Multiple visual styles (e.g., realism, sketch, abstraction, illustration, painting, khokhloma, logos, anime, pixel art, portrait, cyberpunk)
• Negative prompt support (what should not appear in the image or video)
• PNG export
• Image-to-animation option

Video generation typically takes longer than creating static images, with processing ranging from a few seconds to a couple of minutes.