z-image-turbo-lora

This documentation is valid for the following list of our models:

alibaba/z-image-turbo-lora

Try in Playground

Model Overview

An ultra-fast 6B-parameter text-to-image model with LoRA support (see a separate example of how to use it).

Setup your API Key

If you don’t have an API key for the AI/ML API yet, feel free to use our Quickstart guide.

API Schema

post

Body

modelstring · enumRequiredPossible values:

promptstring · max: 4000Required

The text prompt describing the content, style, or composition of the image to be generated.

image_sizeany ofOptionalDefault: landscape_4_3

string · enumOptional

The size of the generated image.

Possible values:

output_formatstring · enumOptional

The format of the generated image.

Default: pngPossible values:

enable_prompt_expansionbooleanOptional

If set to True, prompt will be upsampled with more details.

Default: true

num_inference_stepsinteger · min: 1 · max: 8Optional

The number of inference steps to perform.

seedinteger · min: 1Optional

The same seed and the same prompt given to the same version of the model will output the same image every time.

num_imagesnumber · min: 1 · max: 4Optional

The number of images to generate.

Default: 1

enable_safety_checkerbooleanOptional

If set to True, the safety checker will be enabled.

Default: true

accelerationstring · enumOptional

The speed of the generation. The higher the speed, the faster the generation.

Default: regularPossible values:

Responses

200Success

application/json

post

/v1/images/generations

async function main() {
  const response = await fetch('https://api.aimlapi.com/v1/images/generations', {
    method: 'POST',
    headers: {
      'Authorization': 'Bearer <YOUR_AIMLAPI_KEY>',
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      model: 'alibaba/z-image-turbo-lora',
      prompt: 'A T-Rex relaxing on a beach, lying on a sun lounger and wearing sunglasses.',
    }),
  });

  const data = await response.json();
  console.log(JSON.stringify(data, null, 2));
}

main();

200Success

{
  "data": [
    {
      "url": "https://cdn.aimlapi.com/generations/hedgehog/1749730923700-29fe35d2-4aef-4bc5-a911-6c39884d16a8.png",
      "b64_json": null
    }
  ],
  "meta": {
    "usage": {
      "tokens_used": 120000
    }
  }
}

Example #1: Standard Text-to-Image

Let's generate an image of the specified size using a simple prompt.

import requests
import json  # for getting a structured output with indentation

def main():
    response = requests.post(
        "https://api.aimlapi.com/v1/images/generations",
        headers={
            # Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
            "Authorization": "Bearer <YOUR_AIMLAPI_KEY>",
            "Content-Type": "application/json",
        },
        json={
            "model": "alibaba/z-image-turbo-lora",
            "prompt": "A T-Rex relaxing on a beach, lying on a sun lounger and wearing sunglasses.",
            "image_size": {
                "width": 1440,
                "height": 512
            }
        }
    )

    data = response.json()
    print(json.dumps(data, indent=2, ensure_ascii=False))

if __name__ == "__main__":
    main()

async function main() {
  const response = await fetch('https://api.aimlapi.com/v1/images/generations', {
    method: 'POST',
    headers: {
      // Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
      'Authorization': 'Bearer <YOUR_AIMLAPI_KEY>',
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      model: 'alibaba/z-image-turbo-lora',
      prompt: 'A T-Rex relaxing on a beach, lying on a sun lounger and wearing sunglasses.',
      image_size: {
        width: 1440,
        height: 512
      },
    }),
  });

  const data = await response.json();
  console.log('Generation:', data);
}

main();

Response

{
  "data": [
    {
      "url": "https://cdn.aimlapi.com/flamingo/files/b/0a84de54/-IaUBYEQiYqRaeT7oZvus.png"
    }
  ],
  "meta": {
    "usage": {
      "tokens_used": 17850
    }
  }
}

We obtained the following 1440x512 image by running this code example:

Example #2: Text-to-Image with LoRA Fine-Tuning

The alibaba/z-image-turbo-lora model supports applying up to three LoRA adapters to modify the style or behavior of the base model.

The loras parameter is an array of objects, not strings. Each object describes a single LoRA:

path — where to load the LoRA from (Hugging Face repo ID, direct URL to the weights file, or local path).
scale — how strongly this LoRA should influence the result (typically between 0.6 and 1.0).

Community LoRAs on Hugging Face that are compatible with Z-Image Turbo

Style LoRAs

renderartist/Classic-Painting-Z-Image-Turbo-LoRA – classic oil painting / “old masters” look, museum-like style.
renderartist/Coloring-Book-Z-Image-Turbo-LoRA – “coloring book” style with clear outlines and flat fills, good for children’s illustrations and icons.
ostris/z_image_turbo_childrens_drawings — a stylish LoRA adaptation for generating artistic children's drawings.
renderartist/Technically-Color-Z-Image-Turbo – vivid, cinematic Technicolor-style colors and dramatic lighting.
suayptalha/Z-Image-Turbo-Realism-LoRA – boosts photorealism; includes the trigger word Realism.
AlekseyCalvin/HistoricColor_Z-image-Turbo-LoRA – historical early-1900s color photography inspired by three-color processes.
MGRI/Z-Image-Turbo-Panyue-Lora – character LoRA for the digital persona Panyue (triggers: Panyue, 潘悦).

Technical / utility LoRAs

GuangyuanSD/Z-Image-Re-Turbo-LoRA – Re-Turbo adapter that restores Turbo-level speed while behaving like a de-turbo model for training and advanced workflows.
ostris/zimage_turbo_training_adapter – training adapter / de-distill LoRA used as a base for building new LoRAs on top of Z-Image Turbo.

Let's generate an image using a LoRA to influence the visual style.

import requests
import json  # for getting a structured output with indentation

def main():
    response = requests.post(
        "https://api.aimlapi.com/v1/images/generations",
        headers={
            # Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
            "Authorization": "Bearer <YOUR_AIMLAPI_KEY>",
            "Content-Type": "application/json",
        },
        json={
            "model": "alibaba/z-image-turbo-lora",
            "prompt": "A T-Rex relaxing on a beach, lying on a sun lounger and wearing sunglasses.",
            "image_size": {
                "width": 1440,
                "height": 512
            },
            "loras":[
                {
                  "path": "ostris/z_image_turbo_childrens_drawings",
                  "scale": 0.85
                }
            ]
        }
    )

    data = response.json()
    print(json.dumps(data, indent=2, ensure_ascii=False))

if __name__ == "__main__":
    main()

async function main() {
  const response = await fetch('https://api.aimlapi.com/v1/images/generations', {
    method: 'POST',
    headers: {
      // Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
      'Authorization': 'Bearer <YOUR_AIMLAPI_KEY>',
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      model: 'alibaba/z-image-turbo-lora',
      prompt: 'A T-Rex relaxing on a beach, lying on a sun lounger and wearing sunglasses.',
      image_size: {
        width: 1440,
        height: 512
      },
      loras:[
        {
          path: 'ostris/z_image_turbo_childrens_drawings',
          scale: 0.85
        }
      ]
    }),
  });

  const data = await response.json();
  console.log('Generation:', data);
}

main();

Response

{
  "data": [
    {
      "url": "https://cdn.aimlapi.com/flamingo/files/b/0a85ed1b/j-z8clom9AqL_2ZTsGygk.png"
    }
  ],
  "meta": {
    "usage": {
      "credits_used": 17850
    }
  }
}

We obtained the following 1440x512 image by running this code example:

The scale parameter significantly affects the output. The previous example was generated with scale = 0.85, and here’s what happens when we increase it to 1.0:

Previousz-image-turbo NextByteDance

Last updated 16 days ago

Was this helpful?