GPT Image 2.5 Flare

GPT Image 2.5 Flare is an OpenAI image model that OpenAI describes as its fastest for high-quality, everyday image generation. It accepts text and image inputs, produces images, and supports generation and editing, including inpainting. Masonry calls it through OpenAI's API.

OpenAI positions Flare as its fastest model for everyday generation and Sunburst as its most capable model for generation and editing, recommended where editing precision matters most. Both share the same quality levels, sizes, and inputs in Masonry, and Masonry charges both the same at the same quality and size. OpenAI does not publish more detail on what the names Flare and Sunburst mean.

OpenAI says the GPT Image 2.5 models add xhigh and max quality settings, while earlier GPT Image models stop at high. In Masonry, Flare also exposes a background setting for transparent output. Both models share OpenAI's custom-size rule and accept up to 10 input images in Masonry.

Any size where width and height are multiples of 16, the long edge is at most 3840 pixels, the ratio is between 1:3 and 3:1, and the total is between 655,360 and 8,294,400 pixels. Masonry offers 1K, 2K, and 4K presets for nine common ratios. OpenAI notes that sizes above 2560x1440 are experimental.

Low, medium, high, xhigh, and max. Masonry defaults to low, which is the fastest and cheapest setting. Higher settings use more output tokens and take longer, so they cost more credits per image.

Yes. Add one or more input images, up to 10 in Masonry, and the request goes to OpenAI's edit endpoint. You can add a mask to mark the area to change. With several input images, the mask applies to the first one, and OpenAI notes that masking is prompt-based, so describe the change in the prompt as well.

Yes. Masonry accepts up to 10 input images per request, which the model can edit or use as references. For work where a real product, person, or logo must stay exactly faithful, OpenAI recommends Sunburst for precise editing.

Yes. Set the background to transparent and the output is a PNG with no backdrop. The other background options are auto and opaque.

One to four per request in Masonry. Generating several at once is a quick way to compare directions before choosing one to refine.

It can render short, exact text well when the prompt quotes it. In Masonry's own use, an editorial cover with typeset text was accepted on the first try, and in our seven-line coffee-label test (October 31, 2026) it rendered every line exactly, including punctuation, a date, and a lowercase URL. OpenAI notes the model can still struggle with precise text placement and clarity, so check every word on dense layouts, quote each string, and regenerate with corrected strings rather than editing small type.

Yes. New Masonry accounts start with free credits, and Flare is the image model free accounts generate with. Masonry bills in credits by quality and size, and the editor shows the exact cost before you generate.

Write a clear description of subject, setting, composition, lighting, and style, and put exact on-image words in quotation marks with their position. Start at low quality to explore, then raise quality on the image you keep. For anything that must stay faithful to a real asset, or needs a precise edit, switch to Sunburst.

Open a Masonry canvas, pick GPT Image 2.5 Flare from the model selector, and describe the image you need: a product shot, an ad creative, a social post. Masonry generates it, then you refine, edit, and combine GPT Image 2.5 Flare with other models in one workspace.

Yes. New Masonry accounts get 50 free credits for GPT Image 2.5 Flare at Standard quality, with no card required. Subscribe to unlock higher quality levels and every other model.

Describe the image in plain, specific language, naming the subject, setting, composition, lighting, and style. For any words that must appear in the image, put each exact string in quotation marks and say where it goes. When a dense layout comes back with a small text error, regenerate with the corrected, fully quoted strings rather than editing the image, because an edit pass on small type can introduce new errors.

GPT Image 2.5 Flare is built by OpenAI. Inside Masonry it runs alongside 50+ image and video models, so your team can pick the right one for each brief without switching tools.