Image generation

Image generation API

Generate and edit images with GPT Image and Grok Imagine through the OpenAI-compatible images endpoint, using the same key as every other model.

Models

ModelBilledParameters
gpt-image-2, gpt-image-2.5-flare, gpt-image-2.5-sunburstBy tokens, so size and quality change the pricesize, quality, response_format
grok-imagine-image, grok-imagine-image-qualityA fixed price per pictureaspect_ratio, resolution, response_format

Prices are on each model page. All of them answer on POST /v1/images/generations with Authorization: Bearer, so switching is a change to the model field plus that family’s parameters.

Generate an image

export TOKEN_SHARE_KEY="<your-pool-api-key>"

curl -sS "https://token-share.app/v1/images/generations" \
  -H "Authorization: Bearer $TOKEN_SHARE_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-image-2","prompt":"A cute baby sea otter","size":"1024x1024","quality":"high","response_format":"b64_json"}'

response_format: "b64_json" returns the picture inline as base64; "url" returns a link to download it. With the OpenAI SDK, pass the Grok parameters through extra_body.

Edit an image

POST /v1/images/edits takes a multipart form: the source image, a prompt, and the model. It is checked against the same model list as generation.

export TOKEN_SHARE_KEY="<your-pool-api-key>"

curl -sS "https://token-share.app/v1/images/edits" \
  -H "Authorization: Bearer $TOKEN_SHARE_KEY" \
  -F model="gpt-image-2" \
  -F image="@input.png" \
  -F prompt="Add a small red hat"

/v1/images/variations is not supported; use /v1/images/edits. Video is a separate model and route — see the Grok Imagine video guide.

What it cost

GPT Image responses include a usage object with the tokens you were billed for. For any image request, pass the x-token-share-request-id response header to the Billing API to read the exact charge.

Ready? Three steps to start

  1. 1Sign in to the console
  2. 2Top up your balance
  3. 3Create an API key
Get an API keyView pricing