GPT Image 2.5
OpenAI's newest image model, for generation and scoped edits
Generate photoreal images and typographic layouts, or edit an existing image so only the part you named changes while the subject, composition, and background stay put.
Made with GPT Image 2.5
Live playground
Try GPT Image 2.5 in your browser
Generate or edit an image with your imageat account, then use the same model through the API.
Generate from text
Turn natural-language art direction into finished images.
Edit with references
Guide composition, subject, identity, or style with existing images.
Production output
Choose resolution, aspect ratio, quantity, and output format per request.
Build image workflows with GPT Image 2.5
Generate photoreal images and typographic layouts, or edit an existing image so only the part you named changes while the subject, composition, and background stay put. Use the playground for quick visual testing, then copy the JSON request into a server-side integration with the same settings.
- Text rendering that holds up in posters, packaging, and UI
- Edits scoped to the instruction across many rounds
- Low through Max quality tiers, priced per tier
- Transparent-background cut-outs straight from the model
Ready to integrate GPT Image 2.5?
Create an API key and send your first image request with the same settings you tested above.