Kandinsky AI image bot
In short
An image maker for people who would rather skip English prompts. You describe the scene in Russian, the way you would explain it to a friend, and a picture comes back. The quality is decent, though the top Western models still have the edge.
About the bot
Kandinsky is the image model Sber keeps developing. This bot was one of the first ways to try it on Telegram, back in the 2.1 days, which is where the number in its handle comes from. The model has been updated several times since.
Using it is simple. A button sets the mode, then you type a prompt or upload a picture of your own to rework. Prompts can be written in Russian, so there is no need to translate or hunt for the right English keywords.
We liked it for greeting cards, illustrations for posts and silly scenes to drop into a conversation. There are plenty of styles, from watercolour to something close to a photo. For a blog cover or a temporary avatar the output is good enough.
The weak spots are the usual ones for generators: hands, small lettering and busy scenes with several people work maybe half the time. Expect to regenerate a few times. If you need artwork for a real ad campaign, you will want a stronger paid tool.
You'll like it if
- you want to generate images from Russian prompts
- you need pictures for posts, cards and messages
- you do not want to pay per generation
Skip it if
- you need exact lettering or realistic hands
- you produce commercial design with strict standards
Your rating
No votes yet — be the first.Tap a star — you can change your vote later.
Alternatives
Similar games and bots we recommend