GPT Image 2.6 Sunburst API: Image Generation and Editing

Overview

GPT Image 2.6 Sunburst is an asynchronous API for text-to-image generation and reference-guided image editing. Each successful task returns one file.

Capability	Value
Model ID	`gpt-image-2.6-sunburst`
Modes	`text-to-image`, `image-to-image`
Output formats	PNG, JPEG, and WebP in both modes. GIF and SVG in `text-to-image`.
Reference images	1–16 ordered URLs in `image-to-image`
Edit guide	One optional mask or markup on the first reference image
Resolution tiers	`1k`, `2k`, `4k`
Quality tiers	`low`, `medium`, `high`, `xhigh`, `max`

Endpoints and authentication

Base URL:

https://api.apixo.ai/api/v1

Method	Endpoint	Purpose
`POST`	`/generateTask/gpt-image-2.6-sunburst`	Create an image task
`GET`	`/statusTask/gpt-image-2.6-sunburst?taskId={taskId}`	Read task status and results

Use Authorization: Bearer YOUR_API_KEY and Content-Type: application/json.

Quickstart

Generate a GIF:

curl -X POST "https://api.apixo.ai/api/v1/generateTask/gpt-image-2.6-sunburst" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input":{"mode":"text-to-image","prompt":"A paper boat rocking gently in a seamless loop","output_format":"gif","frame_count":9}}'

For SVG, set output_format to svg. To edit a reference image using an annotated guide, submit this request body:

{
  "input": {
    "mode": "image-to-image",
    "prompt": "Replace the circled object with a blue vase.",
    "image_urls": ["https://example.com/room.png"],
    "markup_url": "https://example.com/markup.png"
  }
}

Parameters

string

default:"async"

Use async to poll for results or callback for webhook delivery.

string

Public HTTPS URL that receives the final task payload. Required when request_type is callback. See Webhooks.

object

required

GPT Image 2.6 Sunburst input parameters.

Show properties

string

required

Generation mode. Use text-to-image or image-to-image.

string

required

Non-blank prompt describing the image or edit.

string[]

Ordered public HTTP(S) reference image URLs. Required for image-to-image. Supply 1–16 URLs.

string

Optional public HTTP(S) mask URL for the first reference image. For image-to-image only. White or transparent regions indicate where to edit. Supply either mask_url or markup_url.

string

Optional public HTTP(S) URL of a transparent PNG marking the first reference image. For image-to-image only. Use arrows, circles, or text to guide the edit. Supply either markup_url or mask_url.

string

Output aspect ratio. Use 1:1, 1:2, 2:1, 1:3, 3:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 9:21, or 21:9. When omitted, text-to-image uses 1:1 and image-to-image follows the first reference image.

string

default:"1k"

Output resolution tier. Use 1k, 2k, or 4k. Affects pricing.

string

default:"medium"

Output quality tier. Use low, medium, high, xhigh, or max. Affects pricing.

string

default:"high"

Planning effort. Use low, medium, or high.

string

default:"png"

Use png, jpeg, or webp in either mode. text-to-image also accepts gif and svg.

integer

default:"9"

Number of GIF frames, from 2 through 16. Only used for GIF output. Affects the GIF price.

integer

default:"200"

Duration of each GIF frame, from 20 through 2,000 milliseconds. Only used for GIF output.

Poll for results

The create response returns data.taskId. Save it and use it to query the task:

curl "https://api.apixo.ai/api/v1/statusTask/gpt-image-2.6-sunburst?taskId=YOUR_TASK_ID" \
  -H "Authorization: Bearer YOUR_API_KEY"

When data.state is success, decode data.resultJson as JSON and read resultUrls:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "YOUR_TASK_ID",
    "state": "success",
    "resultJson": "{\"resultUrls\":[\"https://example.com/result.gif\"]}"
  }
}

Continue polling while state is processing or processing_r2. When state is failed, read failCode and failMsg when present. With request_type: "callback", use the supplied webhook URL to receive the terminal result. See Webhooks.

Billing

For PNG, JPEG, and WebP, use the base price for the requested mode, quality, and resolution. SVG adds $0.020 to the text-to-image base price. A GIF of 2–16 frames costs the same as a static image at the same quality and resolution. The image-to-image base price includes the first reference image. Each additional reference image costs $0.015, and a mask or markup adds $0.015 once. Complexity and frame duration do not affect the standard price. At medium and 1k, PNG and GIF text-to-image each cost $0.050, and SVG costs $0.070. An edit with three reference images and one markup costs $0.110. These are standard prices before discounts. See the full pricing table and credit conversion and rounding.

Latency and polling

Allow 3–5 seconds between status requests. For production workloads, use callback delivery.

MCP integration

The complete model schema declares the result format in mcp.result and attachment paths in mcp.attachments. Bind files to image_urls, mask_url, or markup_url and preserve the order of reference images. The API validates mode requirements and the choice of one edit guide.

​Overview

​Endpoints and authentication

​Quickstart

​Parameters

​Poll for results

​Billing

​Latency and polling

​MCP integration

​Related links

Overview

Endpoints and authentication

Quickstart

Parameters

Poll for results

Billing

Latency and polling

MCP integration

Related links