Z-Image Turbo is positioned in the MakeInfluencer registry for sub-second photorealistic image generation. That makes it a useful model for concept exploration, thumbnail drafts, ad-variation testing, and interactive products where latency matters more than using the most expensive image engine for every attempt.

## Current API Inputs

| Field          | Options                                             |
| -------------- | --------------------------------------------------- |
| `prompt`       | Required generation instruction                     |
| `size`         | Fixed square, landscape, and portrait pixel presets |
| `imageUrl`     | Optional source image for image-guided output       |
| `strength`     | 0–1 transformation strength, default 0.6            |
| `outputFormat` | `jpeg`, `png`, or `webp`                            |

Use the [live Z-Image Turbo API page](/developers/models/z-image-turbo-api) for exact presets and current credit cost.

## Fast Text-to-Image Example

```bash
curl --request POST \
  https://www.makeinfluencer.ai/api/v1/models/z-image-turbo/preview-credits \
  --header "Authorization: Bearer $MAKEINFLUENCER_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "input": {
      "prompt": "Clean ecommerce portrait of a streetwear creator, soft gray background, direct flash",
      "size": "720*1280",
      "strength": 0.6,
      "outputFormat": "webp"
    }
  }'
```

Once approved, send the same request to `/models/z-image-turbo/generations` with an idempotency key.

## Image-Guided Variation

Add `imageUrl` when the workflow starts from an existing image. The `strength` field controls how far the result can move from that source.

- Lower strength favors the source composition
- Higher strength allows a more aggressive transformation
- A mid-range default is a sensible starting point for variation testing

Always review identity and product details. Strength is not a guarantee that a face, logo, or package will remain exact.

## High-Iteration Product Pattern

<StepFlow>
  <Step
    number="1"
    title="Draft Quickly"
    description="Generate several low-friction concepts with Z-Image Turbo using a stable prompt template."
  />
  <Step
    number="2"
    title="Collect Signals"
    description="Let a person, creative score, or experiment choose the strongest direction."
  />
  <Step
    number="3"
    title="Refine Selectively"
    description="Edit the winner with Qwen or regenerate it on a premium model only when the use case requires it."
  />
  <Step
    number="4"
    title="Animate the Winner"
    description="Spend video credits only after the image passes identity and campaign review."
  />
</StepFlow>

This funnel protects gross margin. A product does not need to use its highest-cost model for every idea a user abandons.

## Choosing an Output Format

| Format | Use it for                                            |
| ------ | ----------------------------------------------------- |
| JPEG   | Small photographic previews and feeds                 |
| WebP   | Web delivery when modern format support is acceptable |
| PNG    | Lossless assets, graphics, or another editing stage   |

Choose the format based on the next consumer in the pipeline, not habit. A preview grid may benefit from WebP while an editing workflow may prefer PNG.

## Z-Image Turbo vs Dreamina V3

Use Z-Image Turbo when iteration speed and fixed practical output sizes are the priority. Use [Dreamina V3](/developers/models/dreamina-v3-api) when you want its broader aspect-ratio workflow and a clear optional single-reference editing mode.

<Callout type="info" title="Latency Is a Product Feature">
  Faster drafts can improve activation because users reach a first usable result
  before abandoning the workflow. Reserve premium models for stages where the
  quality difference can increase conversion or selling price.
</Callout>

## API FAQ

### Is Z-Image Turbo really sub-second?

The current MakeInfluencer model registry lists typical inference below one second. End-to-end API time also includes queueing, network, storage, and status updates, so do not hard-code a sub-second SLA.

### Can it use an existing image?

Yes. Supply `imageUrl` and tune `strength` between 0 and 1.

### How many images does one request return?

The model capability is one output per underlying generation. The generic image request can queue multiple copies, each tracked by its returned generation ID.

### How do I avoid duplicate charges on retry?

Send a stable, unique `Idempotency-Key` with the generation request. Replaying the same key and payload returns the first response.


Use the Z-Image Turbo API for fast text-to-image and optional image-guided generation with fixed output sizes, strength control, and JPEG, PNG, or WebP.

Z-Image Turbo API Guide: Fast AI Image Generation

How do I avoid duplicate charges on retry?

Z-Image Turbo API Guide: Fast AI Image Generation

Create Your AI Influencer

Continue Reading

AI Image Editing Guide

Seedance 2 Fast API Guide: Faster Video with Audio

Best AI Video Generators 2026

Ready to try it yourself?