Meet the team that turns your ideas into work.

Discover EightBots
EIGHTBOTS
Contact salesSign up
Kandinsky LabKandinsky 2

Kandinsky 2.1

Kandinsky 2.1 combines an image prior with a latent image decoder. It supports text-driven image generation and mixing of image and text conditions.

Distinguish the generator from its encoders

CLIP image representations connect the prior and decoder. The package also contains MoVQ image encoding and decoding weights.ai-forever/Kandinsky_2.1 — model card and files

Specifications & access

Kandinsky 2.1 specifications
DetailValue and conditions
AccessDownloadable
Model identifiersai-forever/Kandinsky_2.1kandinsky-community/kandinsky-2-1
Diffusion U-Net1.22B parametersai-forever/Kandinsky_2.1 — model card and files
Weight licenseApache 2.0ai-forever/Kandinsky_2.1 — model card and files