GPT Image 2

Official Group · model: gpt-image-2 · Token billing

This page covers API keys created in the Official Group. The key selects the group, so keep the public model ID gpt-image-2 in every request and do not add a group field.

At a glance

Item	Official Group behavior
API key	Must belong to the Official Group
Public model ID	`gpt-image-2`
Submission	Asynchronous, returns `202 Accepted`
Billed on	The token `usage` returned with the completed task
Pricing	Official Group token rates on the model list

Generate an image

APIPOOL_ENDPOINT="https://app.apipool.dev"

curl "$APIPOOL_ENDPOINT/v1/images/generations" \
  -H "Authorization: Bearer $APIPOOL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "A studio product photo of a blue glass sphere",
    "size": "1:1",
    "resolution": "1k",
    "quality": "low",
    "n": 1
  }'

Edit an uploaded image

Use multipart/form-data for file uploads:

curl "$APIPOOL_ENDPOINT/v1/images/edits" \
  -H "Authorization: Bearer $APIPOOL_API_KEY" \
  -F "model=gpt-image-2" \
  -F "prompt=Change the object to emerald green" \
  -F "image=@./source.png" \
  -F "size=1:1" \
  -F "resolution=1k" \
  -F "quality=low" \
  -F "n=1"

Request parameters

Field	Required	Values and meaning
`model`	Yes	Always `gpt-image-2`
`prompt`	Yes	Image generation or editing instruction
`image`	For file editing	One uploaded source image on `/v1/images/edits`
`size`	No	Output aspect ratio, for example `1:1`
`resolution`	No	Output resolution: `1k`, `2k`, or `4k`
`quality`	No	`auto`, `low`, `medium`, or `high`
`n`	No	Number of requested images, `1`–`4`; default `1`

size sets the aspect ratio only — pick the resolution tier with resolution (1k, 2k, or 4k). The image endpoints reject stream: true.

Submission and polling

The submission returns a task rather than the finished image:

{
  "id": "imgtask_01...",
  "object": "image_generation.task",
  "status": "submitted",
  "model": "gpt-image-2",
  "created_at": 1787068800
}

Poll the task using the same API key:

TASK_ID="imgtask_01..."

curl "$APIPOOL_ENDPOINT/v1/tasks/$TASK_ID" \
  -H "Authorization: Bearer $APIPOOL_API_KEY"

See Asynchronous image tasks for all status values, polling rules, temporary result URLs, and retryable responses.

Completed task and usage

Once the billing data is confirmed, Official Group results include standardized token usage:

{
  "id": "imgtask_01...",
  "object": "image_generation.task",
  "status": "completed",
  "model": "gpt-image-2",
  "created_at": 1787068800,
  "completed_at": 1787068842,
  "data": [
    {
      "url": "https://objects.example/results/0.png?...",
      "expires_at": 1787072442
    }
  ],
  "result_expires_at": 1787155242,
  "usage": {
    "input_tokens": 18,
    "input_tokens_details": {
      "text_tokens": 18,
      "image_tokens": 0
    },
    "output_tokens": 196,
    "output_tokens_details": {
      "image_tokens": 196
    },
    "total_tokens": 214
  }
}

Prices are locked at submission time, and the task is charged separately on text input, image input, cached input, and image output tokens. The usage returned with the final task is what you can reconcile against. Current rates are shown under the Official Group on the model list; this page does not repeat prices that can change.

To be charged per delivered image instead, use a key from the Discount Group.