For developers
GetGPTImage is a web app. If you need GPT Image inside your own product, call an image API; here is what to know before you start.
The GPT Image 2 API lets developers generate and edit images in code with the model IDs gpt-image-2, gpt-image-2.5-sunburst and gpt-image-2.5-flare. OpenAI bills them by tokens: $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens. Aiduct (aiduct.ai), the gateway GetGPTImage runs on, serves all three behind one API key.
Add image generation or photo editing to your own app or site.
Create catalog images, thumbnails or variants by the hundred.
Connect image generation to your CMS, store or design workflow.
Request a transparent background for stickers, icons and overlays.
Create a key with OpenAI, or at aiduct.ai to reach all three models through one gateway.
Send the model ID, the prompt, a size and any reference images to the images generation endpoint.
Save the returned image in your own storage before you show it to users.
The three models take the same request parameters. The ones you will set most often:
OpenAI bills all three models by tokens: $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens at the time of writing. Bigger images and higher quality produce more output tokens. A 1024x1024 GPT Image 2.5 image costs about $0.006 at low quality, $0.013 at medium and $0.053 at high.
On Aiduct, GPT Image 2 is listed at $0.087 per image and GPT Image 2.5 is billed at the reported upstream cost. Check the provider's current price list before you build, and prototype prompts in the web app first so experiments do not run up an API bill.
Choose the API when images must be made inside your software, in batches or automatically. Choose the GetGPTImage web app when a person makes the images: there is nothing to build, you refine results by hand, and you pay in credits per image. OpenAI serves its API only in the countries it supports.
No. GetGPTImage is a web app for making images in the browser. For programmatic access use OpenAI's API, or Aiduct, which serves the same three models.
gpt-image-2. The GPT Image 2.5 versions are gpt-image-2.5-sunburst and gpt-image-2.5-flare, all three with the same request parameters.
OpenAI charges $30 per million image output tokens, plus input tokens. On Aiduct, GPT Image 2 is $0.087 per image at the time of writing.
Yes. Send reference images with the prompt and describe the change. Aiduct accepts up to 16 references per request on these models.
Yes. Sign up on GetGPTImage and try all three models in the browser, from 16 credits per 1K image, before you integrate an API.
Try GPT Image 2 and 2.5 in the browser and keep the prompts that work.
Last updated October 17, 2026