How can I integrate the GPT Image 2.6 API into my website?

Call the images endpoint with GPT-Image-2.6 Flare or GPT-Image-2.6 Sunburst. Keep the key on your server.

This page does that when an API key is set. The browser sends the prompt to this site. The site calls the API. The key never has to live in the page.

The request this page sends

It is a POST to /images/generations. A still image looks like this:

{ "model": "gpt-image-2.6-flare", "prompt": "A clear serum bottle in a splash of blue water.", "size": "1024x1024", "quality": "medium", "complexity": "low", "output_format": "png", "n": 1 }

Swap the model to gpt-image-2.6-sunburst when the next edit should match. Quality is low, medium, or high. Complexity uses the same three words and has no separate price. Sizes need to be multiples of 16. The four this page offers are 1024×1024, 1536×1024, 1024×1536, and 1024×1280. Output can be png, jpeg, webp, gif, or svg.

For a GIF, this page also sends frame_count 9 and frame_duration_ms 200. For an SVG, n stays 1. The call does not use partial-image streaming. The response this page reads is the base64 image on the first result.

Edits

A marked edit is a separate call from a fresh prompt. The mark is a transparent PNG the same size as the first image, and it cannot be combined with a mask. Up to four images can go along. GIF and SVG are generation-only, so leave them out of that call.

If the key is missing, this page says so and does not invent a picture. Official details of the model are in Introducing Images 2.6.