About Kandinsky 2.2
Kandinsky 2.2, from AI Forever, has a look — painterly, expressive, unmistakably art-forward rather than photographic. If you want something that feels illustrated instead of photographed, that’s the draw.
It also handles prompts in languages beyond English well, so it’s worth a look if you’d rather not write everything in English.
4 credits an image here.
Best for
- Artistic, painterly imagery
- Multilingual prompting
- Expressive, illustrative styles
- Creative concept art
Not ideal for
- Strict photorealism (use Imagen 4 or FLUX 1.1 Pro)
- Readable text (use Ideogram V3)
- Free generation (try the free models first)
Example prompts
- “An abstract expressionist painting of a city at night, bold colors”
- “A surreal dreamscape with floating islands, painterly style”
- “A vibrant folk-art illustration of a festival, decorative patterns”
How it runs on ArtifyAI
Kandinsky 2.2 generates at 1024×1024 with 75 inference steps (plus a prior step), which is where its dense, artistic detail comes from.
Frequently asked questions
What is Kandinsky 2.2 known for?
A distinctive artistic, painterly style and strong multilingual prompt support.
Can I prompt in other languages?
Yes, Kandinsky 2.2 has good multilingual support, though English prompts still tend to be the most predictable.
How much does it cost?
It is a paid model at 4 credits per image on ArtifyAI.
Can I use the images commercially?
Yes, you retain full usage rights, including commercial use.
Want to explore other models?
Browse all AI models →