Documentation Index
Fetch the complete documentation index at: https://doc.onetoken.one/llms.txt
Use this file to discover all available pages before exploring further.
Bailian Qwen-Image series
Generate images with Tongyi Qwen Bailian Qwen-Image models. The API supports prompt rewriting and strong Chinese-style semantic understanding.
Basics
- Endpoint:
POST /v1/images/generations
- Auth:
Bearer Token
- Content-Type:
application/json
Model overview
Wan models
| Model | Summary | Output image format |
|---|
wan2.6-t2i **recommended** | Wan 2.6 supports flexible sizing under total-pixel and aspect-ratio constraints (same as wan2.5). | Resolution: total pixels between [12801280, 14401440]; aspect ratio [1:4, 4:1]; format: PNG |
wan2.5-t2i-preview **recommended** | Wan 2.5 preview supports flexible sizing under pixel/aspect constraints (for example 768*2700; versions <=2.2 cap one side at 1400). | |
| wan2.2-t2i-flash | Wan 2.2 flash is ~50% faster than 2.1. | Resolution: width and height both in [512, 1440]; format: PNG |
| wan2.2-t2i-plus | Wan 2.2 pro provides improved stability and success rate vs 2.1. | |
| wanx2.1-t2i-turbo | Wan 2.1 turbo. | |
| wanx2.1-t2i-plus | Wan 2.1 plus. | |
| wanx2.0-t2i-turbo | Wan 2.0 turbo. | |
Qwen models
| Model | Summary | Output image spec |
|---|
| qwen-image-2.0-pro | Qwen image generation/editing Pro series. Stronger text rendering, realism, and semantic following. | Resolution: free width/height, with output dimensions each within 512 to 2048. Default resolution is 2028*2048. Format: PNG. Count: 1-6 images. |
| qwen-image-plus-2026-01-09 | Qwen image generation Plus series, good at diverse styles and text rendering. | Default resolution 1664*928. Optional resolutions/aspect ratios: 1664*928 (default) 16:9, 1472*1104 4:3, 1328*1328 1:1, 1104*1472 3:4, 928*1664 9:16. |
| qwen-image | | |
string (required)
Must be application/json.
Authorization string (required)
Bearer API key auth. Example: Bearer sk-xxxx.
Request body
| Field | Type | Required | Description |
|---|
| model | string | Yes | Model name, e.g. qwen-image-plus or qwen-image-max. |
| input | object | Yes | Image description payload. |
| └─ messages | array[object] | Yes | Message list. |
| └─ └─ role | string | No | Role. |
| └─ └─ content | array[object] | | |
| └─ └─ └─ text | string | | Prompt text. |
| parameters | object | No | Generation controls. |
| └─ negative_prompt | string | No | Negative prompt for undesired content. Supports Chinese/English, max 500 chars (extra chars are truncated). Example: low resolution, low quality, distorted limbs/fingers, oversaturated colors, waxy skin, low face detail, over-smoothing, AI-looking artifacts, chaotic composition, blurry/warped text. |
| └─ prompt_extend | boolean | No | Enable smart prompt rewriting. This optimizes only the positive prompt. true (default): enable rewriting for richer details. false: disable for tighter control. |
| └─ watermark | boolean | No | Add Qwen-Image watermark at bottom-right. Default false. |
| └─ size | string | No | Output size, e.g. 1328*1328, 1024*1024; width and height must both be 512-1440. |
| └─ n | integer | No | Number of images. Range 1-4, default 4. Billed per image; set to 1 for testing. |
| └─ seed | integer | No | Random seed in [0,2147483647]. Same seed can improve consistency, but exact results are still not guaranteed. |
Response
Status: 200 OK
| Field | Type | Description |
|---|
| created | integer | Unix timestamp. |
| data | array | Generated images list. |
| └─ url | string | Download URL. |
| └─ b64_json | string | Base64 image data (when requested). |
| └─ revised_prompt | string | Final AI-optimized prompt. |
Request example
{
"model": "wan2.2-t2i-flash",
"prompt": "A kitten running on grass, cinematic look, natural light and shadow",
"size": "960*1390"
}
Response example
{
"created": 1713833628,
"data": [
{
"url": "https://example.com/qwen_output_image.png",
"b64_json": "",
"revised_prompt": ""
}
]
...
}
Developer tips
- Language understanding: Qwen models are strong at Chinese idioms, couplets, and Chinese aesthetics (e.g. Yueyang Tower, blue-and-white porcelain).
- Prompt optimization: Keep
prompt_extend: true to get richer image details.
