GPT Image 2.5 Flare Image to Image API

POST Base URL: https://api.hiapi.ai /v1/tasks

Copy Page

Submit a text-to-image or image-to-image task, then query it with GET /v1/tasks/:id.

gpt-image-2.5-flare Image to image mode.

Request parameters

</> Generate example code

model string required

Use exactly gpt-image-2.5-flare/image-to-image.

example gpt-image-2.5-flare/image-to-image

input object required

Mode parameters.

image_urls string[] required

Image URLs for editing or reference; required, with 1–16 JPEG, PNG, or WebP images. SVG, GIF, and other formats are unsupported. Each image must be at most 20 megapixels and 30 MB, served from a publicly reachable, directly downloadable http(s) URL; data URIs are also accepted.

prompt string required

Prompt length: 1–20,000 characters; exceeding the character limit returns INVALID_REQUEST. Image-to-image is not subject to the text-to-image token guard.

aspect_ratio enum optional

Output aspect ratio.

default auto enum: auto1:13:22:34:33:416:99:1621:927:1616:279:88:9

resolution enum optional

Output resolution. Price per output image: 1K $0.025, 2K $0.04, 4K $0.06 (default 1K).

default 1K enum: 1K2K4K

background enum optional

Optional background treatment: transparent, opaque, or auto; no default. transparent returns an image with an alpha channel at 1K, 2K, and 4K, and only supports png or webp; combining transparent with jpeg returns a 400. 1K transparent output may have slightly different dimensions from opaque output. For transparent output, explicitly ask to extract the subject and keep the background transparent.

enum: transparentopaqueauto

output_format enum optional

Output image format: png, jpeg, or webp; defaults to PNG. Transparent backgrounds only support png or webp, and combining transparent with jpeg returns a 400. The fallback channel may return PNG.

default png enum: pngjpegwebp

callback object optional

Optional terminal-state callback.

url string required

http(s) URL that receives success or failure terminal notifications.

example https://your-domain.com/hiapi/callback

when enum optional

Notify when the task succeeds or fails.

default final enum: final

Request example

Image to image example

Replace the prompt with your description and replace the reference image URL.

Request body⧉

{
  "model": "gpt-image-2.5-flare/image-to-image",
  "input": {
    "prompt": "Keep the main subject and remove the background.",
    "image_urls": [
      "https://example.com/reference.webp"
    ],
    "aspect_ratio": "auto",
    "background": "transparent",
    "resolution": "1K"
  }
}

Success response

Response body⧉

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "tk-hiapi-..."
  }
}

Reference image requirements

• Pass 1–16 reference images per request; more than 16 returns a synchronous 400 (maxItems).
• PNG, JPEG/JPG, and WebP are supported. SVG is rejected with a synchronous 400; GIF, BMP, TIFF, and other formats fail the task.
• Each reference image must be at most 20 megapixels (about 4472×4472) and 30 MB; limits apply per image.
• URLs must be publicly reachable, directly downloadable http(s) links; login-protected and hotlink-protected URLs are not supported. Data URIs are accepted. A 404, an HTML page, or a corrupted file ends the task with status=fail.

Output pixel size

• Output pixels vary with the resolution tier and aspect ratio; use the actual dimensions in the task output.

Getting the result

1. Receive a taskId after submission.
2. Poll GET /v1/tasks/:id or wait for the terminal callback.
3. When status=success, download the image from data.output[].url before expireAt.
4. When status=fail, inspect data.error, correct the request, and resubmit.