Skip to main content
POST
Image Generation
Use GPT Image 2.5 Flare for everyday generation and editing; Sunburst targets detailed compositions. The 2.6 variants are separate choices with their own prices. Read /api/v2/models?category=image for actual availability.
Put controls inside params, rather than at the request top level. Before generating, calculate the price using the same controls and count. All four models range from 1–18 tokens for one image across current presets; the base price does not cover every combination. SVG, GIF, markup, transparency, partial-image streaming, extra quality controls and tools are outside this phase. Editing uses whole-image references. The old gpt-image-2 remains in history and accepted jobs, and is removed from new selection. A response can be queued / processing; poll GET /api/v2/generations/{id} until completion. Do not create a new paid request just to retrieve the same result again.