Image generation and editing
Select the right model for text-to-image and image editing.
Migrating from closed-source models to Model Studio
If you are using Nano Banana, GPT Image, Midjourney, or Seedream, refer to the following table to select the corresponding Model Studio models.
Capability tier | Representative closed-source models | Recommended Model Studio models |
|---|---|---|
High quality | Nano Banana Pro, GPT Image, Seedream 4.0 |
|
Balanced | Imagen 4 Ultra, Midjourney v7 |
|
Fast and cost-effective | FLUX.2 |
|
Text-to-image
We recommend qwen-image-3.0-pro, the flagship Qwen Image 3.0 model. It supports intelligent agent prompt rewriting, excels at text rendering, and can accurately generate Chinese and English text that matches real physical materials. For detailed usage, see Text-to-image.
When to use wan2.7-image-pro
- You need brand color control with palette functionality
- You need higher resolution —
wan2.7-image-prosupports text-to-image resolutions up to 4096x4096
When to use z-image-turbo
- You only need image generation (no editing functionality)
- Speed or cost is the priority — 10x faster generation at about one-fifth the cost
- Photorealistic portraits and product photos
Image editing
We recommend qwen-image-3.0-pro, which delivers powerful complex content generation and high-fidelity rendering. It supports up to 4.5k tokens of input, can generate complex layouts, supports image-in-image and dense information typesetting, and approaches real photographic quality. For detailed usage, see Image editing.
When to use wan2.7-image-pro
It supports multi-image reference (up to 9 input images), bounding-box interactive editing, and character-consistent multi-image generation.
Recommended models
Model ID | Use cases | Text-to-image | Editing | Max outputs | Max resolution |
|---|---|---|---|---|---|
| Complex layout generation, small text rendering, multilingual fonts | Supported | Supported | 6 | 2048x2048 |
| Same as above, with faster generation | Supported | Supported | 6 | 2048x2048 |
| Text rendering, brand colors, character-consistent multi-image generation, multi-image editing | Supported | Supported | 4 (12 consecutive) | 4096x4096 (text-to-image) / 2048x2048 (editing) |
| Same as above, with faster generation and up to 2K resolution | Supported | Supported | 4 (12 consecutive) | 2048x2048 |
| Fast generation, low cost, photorealistic portraits | Supported | Not supported | 1 | 2048x2048 |
All models
Wan
| Model ID | Text-to-image | Editing | Max outputs | Max resolution |
|---|---|---|---|---|
| Supported | Supported | 4 (12 consecutive) | 4096x4096 (text-to-image) / 2048x2048 (editing) |
| Supported | Supported | 4 (12 consecutive) | 2048x2048 |
| Supported | Supported | 4 | 1440x1440 |
| Supported | Supported | 4 | 1440x1440 |
| Supported | Not supported | 4 | 1440x1440 |
| Not supported | Supported | 4 | 1280x1280 |
| Supported | Not supported | 4 | 1440x1440 |
| Supported | Not supported | 4 | 1440x1440 |
| Supported | Not supported | 4 | 1440x1440 |
| Supported | Not supported | 4 | 1440x1440 |
| Legacy | ||||
| Not supported | Supported | 1 | 1024x1024 |
Qwen Image
Model ID | Text-to-image | Editing | Max outputs | Max resolution |
|---|---|---|---|---|
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
| Supported | Supported | 6 | 2048x2048 |
Z-Image
Model ID | Text-to-image | Editing | Max outputs | Max resolution |
|---|---|---|---|---|
| Supported | Not supported | 1 | 2048x2048 |