POST
https://octopusx.ai
/
v1
/
images
/
generations
Doubao Seedream Image Generation API
curl --request POST \
  --url https://octopusx.ai/v1/images/generations \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "prompt": "<string>",
  "n": 123,
  "size": "<string>",
  "image": [
    "<string>"
  ],
  "response_format": "<string>",
  "quality": "<string>",
  "style": {},
  "background": {},
  "watermark": true
}
'
{
  "created": 1735689600,
  "data": [
    {
      "url": "https://.../images/doubao-img-abc123.png",
      "revised_prompt": "A cute little sea otter floating on the sea under sunlight, cinematic, high-detail"
    }
  ]
}

​
Doubao Seedream Image Generation API

The Doubao Seedream series sends requests through a unified image generation entry point, which is suitable for text-to-image and also supports image-to-image via the image array to pass reference images.
  • The unified entry point is POST /v1/images/generations.
  • Pass the specific Seedream model name in model.
  • Pass the pixel dimension string directly in size, for example 2048x2048 or 2560x1440.
  • image supports image URLs or Base64 arrays, making it suitable for image-to-image and multi-reference-image scenarios.
  • Responses are usually compatible with the OpenAI Images style, commonly url or b64_json.

​
Currently Available Models

  • doubao-seedream-4-0-250828
  • seedream-4-0-250828
  • doubao-seedream-4-5-251128
  • doubao-seedream-5-0-260128

​
Method and Path

POST /v1/images/generations

​
Request Example

curl -X POST https://octopusx.ai/v1/images/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "doubao-seedream-4-0-250828",
    "prompt": "A cute little sea otter floating on the sea under sunlight, cinematic, high-detail",
    "n": 1,
    "size": "2048x2048"
  }'

​
Image-to-Image Example

{
  "model": "doubao-seedream-4-5-251128",
  "prompt": "Preserve the subject and atmosphere of the reference image, and redraw it into a more premium advertising poster",
  "n": 1,
  "size": "2560x1440",
  "image": [
    "BASE64_IMAGE_1",
    "https://example.com/reference-2.png"
  ]
}

​
Response Example

{
  "created": 1735689600,
  "data": [
    {
      "url": "https://.../images/doubao-img-abc123.png",
      "revised_prompt": "A cute little sea otter floating on the sea under sunlight, cinematic, high-detail"
    }
  ]
}

​
Authentication

Authorization: Bearer YOUR_API_KEY

​
Body

​
model
string
required
The Doubao Seedream model name. Common values include doubao-seedream-4-0-250828, doubao-seedream-4-5-251128, and doubao-seedream-5-0-260128.
​
prompt
string
required
Generation prompt. It is recommended to pass this explicitly for both text-to-image and image-to-image.
​
n
integer
Number of images to generate. The Apifox you provided treats this as required; from the existing unified image DTO, this field is an optional pointer type in the gateway. To keep behavior consistent with the current plugin, it is recommended to pass 1.
​
size
string
Output size, passed directly as width and height pixel values. These Doubao models are typically mapped to fixed pixel sizes based on fixed ratios.
​
image
array<string> | string
Optional reference image input. Your Apifox definition describes this as “image URL or BASE64 array”; the current plugin also encodes reference images into a Base64 array and puts them in this field.
​
response_format
string
Optional response format. Common values are url and b64_json. If omitted, the final response format depends on the upstream service and the gateway compatibility layer.
​
quality
string
Quality field. The unified image DTO supports accepting it, but whether Doubao actually consumes this field depends on the final upstream service.
​
style
string | object
Style field. The unified image DTO supports accepting it; whether it takes effect depends on the specific Doubao channel.
​
background
string | object
Background control field. The unified image DTO supports accepting it; whether it takes effect depends on the specific Doubao channel.
​
watermark
boolean
Explicit watermark switch. In the current unified image DTO, this field uses a pointer type, and false is semantically different from omitting it.

​
Size Mapping

Your Apifox is consistent with the plugin’s DOUBAO_SIZE_MAP, so you can use the table below:
Aspect Ratiosize Value
1:12048x2048
4:32304x1728
3:41728x2304
16:92560x1440
9:161440x2560
3:22496x1664
2:31664x2496
21:93024x1296
Unlike Gemini, the interfaces in this Doubao series do not pass a “ratio enum” in your existing plugin. Instead, the ratio is first mapped to a specific pixel size, and that size is then written into size.

​
Text-to-Image vs Image-to-Image

ScenarioKey FieldsDescription
Text-to-imageprompt + sizeThe most basic usage
Image-to-imageprompt + size + imageimage can contain one or more reference images

​
Response

​
created
integer
Generation timestamp.
​
data[].url
string
Returned image URL. When the current plugin encounters this field, it will download the image locally again.
​
data[].b64_json
string
Returned image Base64 data. The current plugin will preferentially use this field.
​
data[].revised_prompt
string
Some upstream services return an adjusted prompt.