# Z-Image AI Image Generator

> The full Z-Image Base model for deliberate final renders, diverse concepts and fine prompt control.

Canonical URL: https://allimageai.com/models/z-image/

Website: All Image AI

Language: English

Z-Image is powered by the full Z-Image Base model by Tongyi MAI. It is the recommended default for richer variation, stronger negative-prompt control and a wider creative search space; Z-Image Turbo remains available as the faster drafting option.

## Model details

- Developer: Tongyi MAI (Alibaba)
- Status on All Image AI: Active
- Last verified: 2026-08-25
- Maximum output: 2K
- Modes: Text to Image, Typography
- Reference images: not supported
- Resolutions: 1K, 2K
- Aspect ratios: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3

## Best for

- Diverse concept exploration
- Bilingual display text
- Negative-prompt control
- 2K final renders

## How to use

1. Leave Z-Image selected for the full Base model.
2. Describe subject, layout and lighting.
3. Add exclusions in Negative Prompt.
4. Compare seeds at 1K, then render the selected direction at 2K.

## Settings explained

### Negative Prompt

Exclude unwanted objects, text, blur or rendering artefacts.

### Steps

Start at 28; use more only when the composition or lettering needs another refinement pass.

## Prompt examples

### Bilingual poster

> Editorial travel poster with the exact title “夜航 NIGHT FLIGHT”, deep navy sky, silver aircraft silhouette, restrained modern typography

### Portrait study

> Natural-light portrait in a quiet bookshop, realistic skin texture, muted film palette, 85mm lens

## Limitations

- Slower than Z-Image Turbo
- Text-to-image only on this exact route
- Higher step counts increase latency

## Pricing and credits

Credits for one image at default quality and mode. Resolution, quality, reference inputs, edit mode and output count can change the total. Check the Generate button before submitting.

| Resolution | Credits / image |
| --- | --- |
| 1K | 1 |
| 2K | 3 |

Failed generations cost 0 credits. Catalog availability does not guarantee runtime availability; the workspace reports whether the selected model can run.

[Credit plans](https://allimageai.com/pricing.md)

## Test results

Testing in progress. No verified benchmark scores have been published for this model.

## Commercial use and image handling

Z-Image Base is released under Apache 2.0; generated use remains subject to applicable law and site terms.

Uploads are processed by the configured inference service. You can delete a generation from your history.

## Frequently asked questions

### How is Base different from Turbo?

Base is slower but more diverse and exposes CFG guidance plus negative prompts; Turbo is the one-credit drafting lane.

### Can it edit an uploaded image?

Not on this exact Base endpoint. Use Z-Image Turbo or another editing model for image-to-image work.

## Release history and sources

January 27, 2026 — Open weights released

This date applies to the event shown. Later API access, weights and availability on All Image AI may have different dates.

- [github.com](https://github.com/Tongyi-MAI/Z-Image)

[Event evidence on the timeline](https://allimageai.com/timeline/#event-z-image)