콘텐츠로 건너뛰기

Open model · Z-Image Turbo

Z-Image Turbo Online: Fast Photoreal Images with Real Text

Z-Image Turbo is a 6B open image model from Tongyi-MAI that renders in 8 steps. It is strong at photoreal scenes and writes English and Chinese text into the image. Run Z-Image Turbo here for 2 credits per image, with nothing to install, or add one photo to remix it.

Try Z-Image Turbo

Text to image

Optional: add one image to repaint it with your prompt.

Ratio
2 credits
Z-Image Turbo close-up portrait of an elderly fisherman in a navy knit cap with harbor boats behind him
Photoreal portrait
Z-Image Turbo street food stall at night with a sign reading LUCKY NOODLE and the Chinese characters 好运面馆
English + Chinese text
Z-Image Turbo flat vector travel poster with a red torii gate, maple leaves and the words KYOTO and AUTUMN 2026
Poster
See every Z-Image Turbo example and prompt
  • 2 credits per image
  • English and Chinese text in the image
  • New accounts get one free image

What Z-Image Turbo is

Z-Image is an open image model family from Tongyi-MAI at Alibaba, released in November 2025. Z-Image Turbo is the distilled member of the family: it needs only 8 function evaluations per image, and the team reports sub-second generation on data-center GPUs and a fit within 16 GB of VRAM on consumer cards. The Z-Image Turbo on this site runs on WaveSpeed's hosted copy.

Developer
Tongyi-MAI (Alibaba)
Released
November 2025: weights on Hugging Face, paper published November 27, 2025
Family
Z-Image (base, 50 steps), Z-Image Turbo (8 steps), plus Z-Image-Omni-Base and Z-Image-Edit
Architecture
6B parameters, single-stream diffusion transformer (S3-DiT)
Steps
8, distilled with Decoupled-DMD and DMDR; guidance scale 0 in the official example
Strengths, per the team
Photorealism, English and Chinese text rendering, and following instructions
License
Apache 2.0
On this site
Text to image at about 1 MP in 7 ratios, or image to image from one photo
Price here
2 credits per image

Z-Image Turbo examples with the exact prompts

Nine Z-Image Turbo runs from November 15, 2026, all at about 1 MP. Each card shows the prompt exactly as sent, one written in Chinese. Use it loads the prompt and ratio into the generator with Z-Image Turbo selected. Each run took 9 to 14 seconds end to end, including the queue.

Z-Image Turbo close-up portrait of an elderly fisherman in a navy knit cap with harbor boats behind him

Photoreal portrait

Z-Image Turbo prompt

Close-up portrait of an elderly fisherman with a weathered face, grey stubble and a navy knit cap, harbor boats in soft focus behind him, overcast daylight, 85mm lens, photorealistic

Skin, stubble and the knit of the cap hold up at 1K, with the harbor kept soft behind him.

Z-Image Turbo street food stall at night with a sign reading LUCKY NOODLE and the Chinese characters 好运面馆

English and Chinese text

Text

Z-Image Turbo prompt

A small street food stall at night with a glowing sign that reads "LUCKY NOODLE" in bold red English letters and the Chinese characters "好运面馆" underneath, neon tubes, wet pavement reflections, light rain, photorealistic

Both lines are correct, LUCKY NOODLE and 好运面馆. Krea 2 Turbo got the Chinese line wrong on the same prompt.

Z-Image Turbo product photo of matte black wireless earbuds in an open case on wet slate

Product shot

Z-Image Turbo prompt

Matte black wireless earbuds in an open charging case on a slab of dark grey slate, small water droplets, soft rim light, commercial product photography, no logo, no text

No logo appeared, and the slate edge and droplets are sharp.

Z-Image Turbo photo of three Japanese souffle pancakes with berries and powdered sugar by a café window

Food

Z-Image Turbo prompt

A stack of three fluffy Japanese souffle pancakes with fresh berries and powdered sugar on a white plate, café table by a window, shallow depth of field, morning light

Exactly three pancakes, as asked, with the window blurred behind.

Z-Image Turbo interior photo of a small kitchen with sage green cabinets, brass handles and herbs on the windowsill

Interior

Z-Image Turbo prompt

Small Scandinavian kitchen in the morning, sage green cabinets with brass handles, white oak countertop, potted herbs on the windowsill, sunlight on the floor, interior photography

Every item from the prompt is there, but the result is plain, closer to a catalog photo than an editorial one.

Z-Image Turbo flat vector travel poster with a red torii gate, maple leaves and the words KYOTO and AUTUMN 2026

Travel poster

Text

Z-Image Turbo prompt

Minimal flat vector travel poster with a red torii gate on a hill and falling maple leaves, cream background, large bold title text "KYOTO" at the top and a small line "AUTUMN 2026" at the bottom

KYOTO and AUTUMN 2026 are spelled correctly, in a clean flat style.

Z-Image Turbo 3D render of a cute clay robot holding a potted cactus on a mint background

3D character

Z-Image Turbo prompt

A cute round robot made of matte clay watering a tiny potted cactus, pastel mint background, soft studio light, 3D render, playful

A clean pastel render on the first try. It holds the cactus rather than watering it.

Z-Image Turbo aerial photo of green rice terraces at sunrise with mist in the valley

Aerial landscape

Z-Image Turbo prompt

Aerial drone photo of green rice terraces curving around a hillside at sunrise, low mist in the valley, a small wooden farmhouse, golden light

The curving terraces, valley mist and low sun read clearly from above.

Z-Image Turbo photo from a Chinese prompt: an orange cat on the wooden windowsill of an old bookshop beside a rainy stone street

A prompt written in Chinese

Chinese prompt

Z-Image Turbo prompt

一只橘猫趴在老书店的木窗台上晒太阳,窗外是下着小雨的石板街,暖色灯光,胶片质感,写实摄影

The prompt asks for an orange cat on the wooden sill of an old bookshop, a rainy stone street outside, warm light and film texture. All of it came through.

Z-Image Turbo image to image

Add one photo and Z-Image Turbo remixes it toward your prompt at the ratio you pick, for the same 2 credits. It keeps the palette and the general scene but redraws the details, so treat it as a remix, not an editor: faces, exact layouts and small objects will change. For precise edits, use Krea 3 or Seedream 5.5.

Rainy neon city street at night with pink and teal signs
Source
Z-Image Turbo image to image result: the neon street turned into a bright snowy winter morning
Result

Night to snowy morning

Image to image

Z-Image Turbo prompt

The same city street on a bright snowy winter morning, fresh snow on the ground and ledges, clear blue sky, people in winter coats, photorealistic

The street canyon and the parked cars carried over; the neon night became a sunny morning in snow.

Concrete interior with a staircase and daylight streaming through tall windows
Source
Z-Image Turbo image to image result: the concrete staircase interior with warm wall lamps added
Result

A partial result

Image to image

Z-Image Turbo prompt

The same concrete interior at night, lit only by warm wall lamps along the staircase, dark windows, calm and moody, architectural photography

It added the warm wall lamps but kept daylight and rebuilt the staircase. This is the limit of a remix: big lighting changes need a real editing model.

Z-Image Turbo compared with Krea 2 Turbo

Two open models, the same prompts, one run each. Credits are what each run costs on this site.

Text in the image

A sign with English and Chinese on it.

A small street food stall at night with a glowing sign that reads "LUCKY NOODLE" in bold red English letters and the Chinese characters "好运面馆" underneath, neon tubes, wet pavement reflections, light rain, photorealistic

Z-Image Turbo street food stall at night with a sign reading LUCKY NOODLE and the Chinese characters 好运面馆

Z-Image Turbo · 2 credits

Both lines correct, including 好运面馆.

Krea 2 Turbo street stall sign: LUCKY NOODLE is correct, the Chinese line reads 好廷面馆

Krea 2 Turbo · 5 credits

LUCKY NOODLE is right; the Chinese reads 好廷面馆, and a second garbled sign appeared. Moodier neon, though.

For signs, posters and packaging with Chinese text, Z-Image Turbo is the model to use here.

The same portrait prompt

A documentary-style portrait of a ceramicist.

Portrait of a ceramicist in her sunlit studio, hands dusted with clay, shelves of unglazed bowls behind her, soft window light, documentary photography

Z-Image Turbo result for the ceramicist portrait prompt

Z-Image Turbo · 2 credits

Clean, bright and literal: centered at the wheel, a tidy studio, little clay on her hands.

Krea 2 Turbo result for the same ceramicist portrait prompt

Krea 2 Turbo · 5 credits

More mood and texture: clay worked into both hands and a stained shirt, backlit by the window.

Z-Image Turbo gives a clean, literal result for less. Krea 2 Turbo costs more and brings more atmosphere.

How to use Z-Image Turbo online

  1. 01

    Write the prompt

    Describe the subject, setting and light in plain English or Chinese. Put any words that should appear in the image in quotes.

  2. 02

    Pick a ratio, or add a photo

    Choose one of 7 ratios; every size is about 1 MP. Add one photo to remix it instead.

  3. 03

    Generate and download

    Each image costs 2 credits and usually lands in about 10 seconds. Results stay in your history to download or reuse.

Z-Image Turbo prompt tips

What worked in the runs above.

  1. 1

    Quote the exact text

    Put each line of text in quotes and say where it goes, as in the sign and the poster. Both English and Chinese came out right.

  2. 2

    Prompt in Chinese if that is easier

    The cat prompt was written only in Chinese and every detail came through. You do not need to translate.

  3. 3

    Add atmosphere on purpose

    Z-Image Turbo is literal: the kitchen came out tidy but plain. Name the light, the film stock or the mood when you want more than a catalog look.

  4. 4

    Count things for it

    "A stack of three pancakes" gave exactly three. Numbers and positions in the prompt are followed closely.

  5. 5

    Iterate cheaply

    At 2 credits an image, run several variations of a prompt and keep the best one.

Z-Image Turbo pricing

Credits are taken when a run starts and refunded if it fails. New accounts get one free image.

RunCredits
Text to imageAbout 1 MP, e.g. 1360 × 768 at 16:9 or 1024 × 1024 at 1:12
Image to imageOne photo, output at the ratio you pick2

Z-Image Turbo FAQ

What is Z-Image Turbo?
Z-Image Turbo is the fast, distilled member of the Z-Image family from Tongyi-MAI. It is a 6B diffusion transformer that renders in 8 steps, released as open weights in November 2025. It is known for photoreal images and for writing English and Chinese text into them.
Is Z Image Turbo free?
The weights are free to download under the Apache 2.0 license, and the team says the model fits in 16 GB of VRAM. On this site there is nothing to set up: each Z-Image Turbo image costs 2 credits, and new accounts get one free image.
Can Z-Image Turbo write Chinese text in images?
Yes. On this page it wrote 好运面馆 correctly on a shop sign, next to an English line, where Krea 2 Turbo got one character wrong. Quote the exact characters in your prompt.
Can I write Z-Image Turbo prompts in Chinese?
Yes. The orange cat example above was prompted entirely in Chinese, and every detail of the prompt appeared in the image.
Is this Z-Image Edit or Z-Image Base?
Neither. This page runs Z-Image Turbo. Its image to image mode remixes a photo toward your prompt; it is not an instruction editor. For edits that keep the rest of the image, use Krea 3 or Seedream 5.5 in the same generator.
Can I use a Z-Image Turbo LoRA here?
Not yet. This generator runs the base Z-Image Turbo checkpoint. LoRAs need the open weights, for example in ComfyUI.
What sizes does Z-Image Turbo make here?
About 1 MP in seven ratios: 1024 × 1024 (1:1), 1184 × 880 (4:3), 880 × 1184 (3:4), 1248 × 832 (3:2), 832 × 1248 (2:3), 1360 × 768 (16:9) and 768 × 1360 (9:16).
Should I use Z-Image Turbo or Krea 2 Turbo?
Use Z-Image Turbo (2 credits) for photoreal, literal images and any Chinese text. Use Krea 2 Turbo (5 credits at 1K) when you want more mood and styling, or a 2K image.

Generate with Z-Image Turbo

Start from any Z-Image Turbo prompt above or a blank box. 2 credits per image, and new accounts get one free image.