Z-Image Turbo
What is Z-Image Turbo?
Z-Image Turbo is Alibaba's January 2026 economy text-to-image model. The pipeline is prompt-only: bilingual typography (Chinese and English), seven aspect ratios (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9), an optional prompt rewriter (off by default; enabling doubles cost), and seed control. No reference images, no editing endpoint. The lowest per-image cost in the Alibaba lineup, built for high-volume prototyping and bilingual creative work where references are not part of the brief.
Key features of Z-Image Turbo
Bilingual in-frame text rendering
Chinese and English typography stays legible inside the image. Useful for retail signage, dual-language packaging, and any bilingual creative work that has to render legibly at thumb size.
Seven aspect ratios from one endpoint
1:1, 2:3, 3:2, 3:4, 4:3, 9:16, and 16:9 cover feed posts, vertical social, landscape banner, and portrait standard. The model does not infer custom crops beyond these.
Lowest cost in the Alibaba image lineup
Per-image price is the lowest of the Alibaba image models. Useful for high-volume prototyping where each iteration is a draft, not a deliverable.
Optional prompt rewriting (off by default)
The `prompt_extend` toggle is off by default; enabling it routes the prompt through Alibaba's rewriter at double the cost. Turn it on selectively when adherence is the bottleneck.
Best for
Rapid prototyping
Many variants per minute for layout approval.
Bilingual packaging
Mixed Chinese and English callouts without bespoke renders.
High-volume social
Scheduled posts that need quick illustrative frames.
Ad iteration
Frequent ratio swaps locked through aspect_ratio presets.
Game UI sketches
Icon and banner first drafts before a slower model pass.
Use cases
Build a high-volume social drop where each frame is a draft with bilingual copy in the same poster; Z-Image Turbo stays cheap enough that the per-call cost is invisible at scale. Run packaging mocks for a Chinese-and-English retail launch where the bilingual layout has to read on shelf without separate masters per language. Sketch game UI banners from text alone before passing the keepers to a slower, higher-quality model. Sweep aspect-ratio variants of an ad creative across the seven presets in a single batch.
API examples
Call Z-Image Turbo from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/image/alibaba/z-image-turbo.
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "alibaba/z-image-turbo",
"input": {
"prompt": "Stylish young woman in front of a cartoon mural",
"aspect_ratio": "3:4"
}
}'
Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.
FAQs
People also ask
Yes. There is no image upload path, so workflows stay prompt based. Move to Qwen Image or another edit-capable model when you need references.
The prompt_extend toggle forwards your prompt through Alibaba's rewriter. It is off by default, and enabling it doubles the cost of the call.
One of the seven presets (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9). The model will not infer a custom crop beyond those.
Z-Image Turbo emphasizes bilingual text rendering. Qwen Image Plus emphasizes stylistic variety. Wan 2.2 Image Flash is the simplest text-to-image surface. Pick by what you need most: text in frame, style range, or simplicity.
