GPT Image 2.5 Flare

GPT Image 2.5 Flare focuses on fast everyday image generation and iterative visual work. It generates and edits images from text and reference inputs, with five explicit quality levels, transparent backgrounds and masked edits. Image input tokens are priced separately from text input tokens.

OpenAI positions Flare for fast everyday generation and repeated visual iteration. It is one of two GPT Image 2.5 models released together. The pair accepts both text prompts and reference images, so the same model can produce an initial image and then revise it using further instructions.

The API offers five explicit quality settings, from low through max, plus automatic selection. Square, portrait and landscape outputs can be requested, and transparent backgrounds are supported. Masked editing is available when a change should concentrate on part of an existing image.

Image and text inputs use separate token prices. Output cost depends on the selected quality, dimensions and actual token consumption; the output-only estimates on this page exclude the prompt and reference images. Inspect the completed generation's cost breakdown for its measured usage.