Seedream 5.5 API for Image Generation and Layer Decomposition

Check the APIXO model catalog for current availability.

Endpoints

Method	Endpoint	Description
POST	`/api/v1/generateTask/seedream-5-5`	Create a generation task
GET	`/api/v1/statusTask/seedream-5-5?taskId=...`	Query task status

Authentication

Use your APIXO API key with Authorization: Bearer YOUR_API_KEY. Send JSON requests to https://api.apixo.ai with Content-Type: application/json.

Request Body

{
  "request_type": "async",
  "input": {
    "mode": "text-to-image",
    "prompt": "A ceramic cup on a transparent background",
    "resolution": "1.5k",
    "aspect_ratio": "auto",
    "background": "transparent"
  }
}

Parameters

Parameter	Type	Rules
`request_type`	string	`async` for polling or `callback` for webhook delivery.
`callback_url`	string	Required when `request_type` is `callback`. Use a public HTTPS URL. See Webhooks.
`input.mode`	string	`text-to-image`, `image-to-image`, or `layer-decomposition`. Default `text-to-image`.
`input.prompt`	string	Required for generation and editing. Optional for layer decomposition.
`input.image_urls`	string[]	Required for image-to-image (1–10 URLs) and decomposition (exactly one URL). Text-to-image ignores this field.
`input.resolution`	string	Accepts `1k`, `1.5k`, or `2k` (uppercase `K` also works). Generation and editing also accept `3k` with a concrete aspect ratio. Decomposition accepts omission for automatic sizing.
`input.aspect_ratio`	string	`auto`, `1:1`, `4:3`, `3:4`, `16:9`, `9:16`, `3:2`, `2:3`, or `21:9`. Use a concrete ratio for `3k`. Decomposition ignores this field and uses automatic aspect ratio.
`input.size`	string	Custom width and height for generation or editing, such as `2880x2880`, `2880X2880`, or `2880*2880`. If provided, it takes precedence over `resolution` and `aspect_ratio`; otherwise both are required. Total pixels must be 921,600–8,294,400 and the width-to-height ratio must be 1:16–16:1. Decomposition ignores this field.
`input.background`	string	`auto` (default), `transparent`, or `opaque`. Transparent output requires PNG. Decomposition ignores this field.
`input.output_format`	string	`png` (default) or `jpeg`. In decomposition this controls the base image format. Other layers are PNG.
`input.watermark`	boolean	Default `false`.
`input.prompt_optimize_mode`	string	Optional. The supported value is `standard`, also the default.
`input.max_images`	integer	For `layer-decomposition`, the maximum total outputs including the base image, 2–100. Default 30. Text-to-image and image-to-image ignore this field and return one image.
`input.text_recognition`	boolean	Decomposition only. Include rich-text recognition when available. Default `false`.

Example

Submit a layer-decomposition task with the following request body.

{
  "request_type": "async",
  "input": {
    "mode": "layer-decomposition",
    "image_urls": ["https://example.com/poster.png"],
    "resolution": "2k",
    "max_images": 10,
    "text_recognition": true
  }
}

Response

Task creation returns data.taskId. Poll the status endpoint or use a webhook for the result. The following example shows a successful decomposition.

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_123",
    "state": "success",
    "resultJson": "{\"resultUrls\":[\"https://example.com/base.png\",\"https://example.com/title.png\"]}",
    "layerDecomposition": [
      {"role": "base", "url": "https://example.com/base.png", "size": "2048x2048", "output_format": "png", "z_index": 0, "background": "opaque"},
      {
        "role": "layer",
        "url": "https://example.com/title.png",
        "size": "800x160",
        "output_format": "png",
        "z_index": 1,
        "background": "transparent",
        "bounding_box": {"absolute": [120, 80, 920, 240]},
        "text_recognition": {
          "text": "Summer Sale",
          "lines": [
            {
              "bounding_box": {"absolute": [0, 0, 800, 160]},
              "runs": [{"text": "Summer Sale", "style": {"bold": true}}]
            }
          ]
        }
      }
    ]
  }
}

Status Response Fields

Field	Description
`taskId`	Task identifier.
`state`	`pending`, `processing`, `success`, or `failed`.
`resultJson`	JSON string containing `resultUrls` in result order.
`layerDecomposition`	Base image followed by the layers. Items may also include `name`, `description`, normalized bounding boxes, and text recognition when available.
`failCode`, `failMsg`	Failure code and message.

Error Codes

Requests can fail validation for missing inputs, out-of-range image dimensions, or incompatible format and background options. A failed task includes failCode and failMsg.

Billing

The initial charge is $0.034 per output for 1k and 1.5k, and $0.045 for 2k. Explicit sizes and 3k with a concrete ratio use the requested pixel count. Decomposition initially charges for up to max_images outputs at the selected rate, or $0.045 per output when resolution is omitted. Final billing uses the actual width × height of each returned image, including the base image and each layer in decomposition. Any overpayment is refunded; the charge does not exceed the initial charge. Text recognition and prompt optimization have no separate fee.

Actual pixels per output	Price per output
Up to and including 2,610,000	`$0.034`
Above 2,610,000 and up to 4,630,000	`$0.045`
Above 4,630,000	`$0.090`

The input price is $0.015 per image, including the first. The total request charge is rounded up to the next $0.001.

MCP integration

The model schema declares mcp.attachments for input.image_urls and mcp.result for decoding image URLs from resultJson.resultUrls. See MCP model contracts.

​Endpoints

​Authentication

​Request Body

​Parameters

​Example

​Response

​Status Response Fields

​Error Codes

​Billing

​MCP integration