Image · prunaai/z-image-turbo
Z-Image Turbo — fast drafts and illustrations from a description
Z-Image Turbo is an ultra-fast 6-billion-parameter text-to-image model from Alibaba’s Tongyi-MAI team; in limko it runs in an optimized version by Pruna AI. A frame costs 0.6 ₽ and is usually ready in a few seconds, queue included. It is the pick when the number of attempts matters more than fine layout: exploring ideas, drafts, illustrations for posts.
from 0.6 ₽ per image
When Z-Image Turbo fits
- Exploring ideas: at this price you can run a dozen wordings and choose a direction.
- Drafts and moodboards before the final on a senior model.
- Illustrations for posts, articles and newsletters that need no text in the frame.
- A short wait: usually a few seconds including the queue, with the image built in 8 generation steps.
- You can write the prompt in your own language, and the generation settings have a switch that translates the prompt into English.
What this model is
Z-Image Turbo was developed by Alibaba’s Tongyi-MAI team, and Pruna AI optimized how it runs. An image is built in just 8 generation steps and without guidance — hence the speed.
In limko the model sits in the image generator and has a single mode — text to image: you cannot attach a photo, and each run gives one variant. To compare several, run the generation again — every run costs the same.
Z-Image Turbo and pricier models
When you need editing, references or lettering, move on to the models in the table.
| Model | What it can do | Price per image |
|---|---|---|
| Z-Image Turbo | text to image only, one variant per run | from 0.6 ₽ |
| GPT Image 2.5 Flare | text in the frame, editing by description | from 3 ₽ |
| Nano Banana 2 Lite | editing and up to five input images, up to four variants per run | from 5.4 ₽ |
| Nano Banana Pro | dense typography, Google’s flagship | from 22.3 ₽ |
Limitations worth knowing
Z-Image Turbo has no negative prompt: the model works without guidance. So describe what should be in the frame, not what to remove: “an empty street” instead of “no people”.
Fine typography and complex scenes with many conditions are not its strong suit. For lettering, take GPT Image 2.5 Flare or Nano Banana Pro; a good draft can be attached to them as a reference and taken to the final.
Prompts for quick drafts
One run, one image, so it is easy to change one detail at a time and compare. Do not ask for text in the frame — other models handle that.
Post illustration
A cozy desk by the window: a laptop, a mug of tea, a stack of books, a potted plant. Soft daylight, warm tones, top-down view.
Cover idea
A lighthouse on a rocky shore in a storm, waves crashing against the rocks, a beam of light cutting through the fog. Dramatic sky, cinematic shot.
Character
A ginger cat in an astronaut suit sits on the Moon and looks at the Earth. Cartoon 3D style, soft light, dark background.
Frequently asked about Z-Image Turbo
How much does a Z-Image Turbo image cost?
0.6 ₽ per frame. Credits are charged only for a successful generation.
Can I upload my own photo?
No, the model works from text only. For editing photos, use P-Image Edit or Nano Banana 2 Lite.
Why doesn’t the negative prompt work?
Z-Image Turbo does not have one: the model works without guidance. Describe in the prompt what you want to see.
Can it draw lettering?
Fine typography is not its strong suit. For text in the frame, GPT Image 2.5 Flare or Nano Banana Pro are more reliable.
Does the model understand my language?
You can write the prompt in your own language; if the result is far from what you had in mind, turn on the switch in the generation settings that translates the prompt into English. No VPN needed.