Why it fits
One open model for production images
Photorealism, bilingual text, and fast turnaround behind a single API key.
Z-Image Turbo
A 6B Apache-2.0 model with strong photorealism and instruction-following.
Bilingual text in images
Reliable English and Chinese typography rendered directly in images, where most models fail.
Fast & low-cost
Roughly 2.5 seconds per image — quick enough to iterate in real time.
OpenAI-compatible
POST /v1/images/generations and get back base64 or a URL with the client you already use.
Pay per image
Per-image pricing across multiple regions, with no idle GPU costs to carry.
Multi-region
Requests route to a healthy region automatically.
Quickstart
Generate an image
Call the OpenAI-compatible images endpoint and get back a base64 image or a hosted URL.
# image generation (synchronous)
image = client.images.generate(
model="z-image-turbo",
prompt="a neon city skyline at dusk",
)
# or call the endpoint directly
POST /v1/images/generations
{ "model": "z-image-turbo", "prompt": "a storefront sign reading hello" }
# -> returns a b64_json image or a hosted URL