GPT Image 2.5

OpenAI's newest image model with two modes in one: Flare for fast everyday generation and Sunburst for precision editing, with medium and high quality tiers.

Resolution: medium, high
Aspect Ratios: 1:1, 16:9, 9:16, 4:3, 3:4, auto

Create with GPT Image 2.5 View Pricing

GPT Image 2.5 Flare sample, golden-hour Lisbon street with a yellow tram, photorealistic

What makes GPT Image 2.5 unique

Two variants in one model: Flare is OpenAI's fastest model for high-quality everyday generation; Sunburst is its most capable for image editing where precision matters most.

Sharper detail and markedly better text rendering than GPT Image 2 — typography, posters, and layouts come out clean.

Roughly 50% lower latency than GPT Image 2 on the Flare variant — everyday images in about 10–20 seconds.

Accepts up to 16 reference images for image-to-image workflows and character-consistent generation.

Medium and high quality tiers so you only pay for the fidelity a job needs.

Supports prompts up to 20,000 characters for detailed creative briefs, style notes, and composition requirements.

Best for

Poster and layout design with real typographyPrecision image editing and retouchingMarketing campaign visualsFast social media creativesProduct concept mockupsReference-guided and character-consistent imagery

How GPT Image 2.5 works

GPT Image 2.5 takes a text prompt and, optionally, up to 16 reference images. Choose the variant — Flare for speed, Sunburst for editing precision — plus a quality tier and aspect ratio, then submit. The model generates the image and returns it in seconds; with references it runs in edit mode, following the supplied subjects and style while applying your prompt.