About these notes

Working notes kept while building a third-party browser interface for OpenAI's GPT Image 2.5. The interface is GPT Image 2.5 Image Generator — it runs both models OpenAI ships, Flare and Sunburst, and prints the model ID the API returned under every result.

The notes exist because most of what I had to work out is not in anybody's documentation: which frames an endpoint will actually accept, what an edit constraint should literally say, how to composite a watermark in a runtime with no imaging library. None of that is proprietary and all of it cost a day each to discover.

Disclosure

I build the product these notes reference, so read them with that in mind. It is an independent interface and is not affiliated with, endorsed by, or connected to OpenAI. Every number here was measured against a live endpoint rather than estimated; where a figure is derived rather than measured, the note says so.

Nothing here is sponsored and there are no affiliate links.