Kandinsky 3
by Sber AI · open-source · Last verified 2026-03-17
Kandinsky 3 is an open-source text-to-image model developed by Sber AI Lab, featuring a novel architecture that combines a diffusion transformer with a high-quality text encoder for strong multilingual prompt support. It delivers competitive image quality with a particular strength in artistic and painterly styles, and is freely available for research and commercial use.
https://ai-forever.github.io/Kandinsky-3/ ↗C
C—Below Average
Adoption: CQuality: B+Freshness: BCitations: C+Engagement: F
Specifications
- License
- Apache 2.0
- Pricing
- open-source
- Capabilities
- text-to-image, image-to-image, inpainting, multilingual-prompts, artistic-styles
- Integrations
- huggingface, diffusers
- Use Cases
- research, artistic-generation, multilingual-workflows, open-source-projects, fine-tuning
- API Available
- Yes
- Parameters
- 3B
- Context Window
- N/A
- Modalities
- text, image
- Training Cutoff
- 2023
- Tags
- image-generation, text-to-image, open-source, sber, russian-ai
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
47.4Adoption
48
Quality
76
Freshness
65
Citations
52
Engagement
0