Turn a prompt into an image or upload a reference to guide an edit. Create with GPT Image 2.6 Flare.
Model overview
GPT Image 2.6 is the model family behind OpenAI’s Images 2.6 release, announced on November 4, 2026. It builds on GPT Image 2.5 with improvements to editing consistency, creative control and generation speed.
OpenAI · Images 2.6The release includes gpt-image-2.6-flare and gpt-image-2.6-sunburst. Compare their official documentation for supported inputs, output settings and availability. This page introduces the official release; site generation availability is shown separately in the studio.
What is new
OpenAI’s release highlights directed edits, more consistent revisions and new ways to take a visual idea beyond a still image.
Give the image a clear subject, setting, light and composition. The Picsart glass-whale example illustrates how an imaginative brief can become a detailed scene. Official example: Picsart / OpenAI.
Mark the part of an image you want to change and describe the edit. The fal room example shows the visual brief before the requested revisions. Official example: fal / OpenAI.
Review the edited room beside its marked-up reference. Keep the approved composition in your brief while changing a particular detail. Official result: fal / OpenAI.
The official release includes GIF workflows and SVG output through the API. The owl is shown here as a static animation poster; open the source to see motion. Official example: OpenAI.
Four-step workflow
Choose GPT Image 2.6 Flare, describe the image, then check your settings and credit quote.
Sign in on Open AI Image and open the image workspace. Describe the subject, setting, composition, lighting, and any required text, or add a reference image that should guide the result.
Check the selected model, output options and credit quote. Generate a first version, then review its composition, text and details against your brief.
With a model that supports image references, describe one change at a time. State which subject, colors or layout should remain consistent.
Download the chosen result from the workspace. You can also take a finished image into a supported image-to-video workflow.
Visual examples
Selected examples from OpenAI’s Images 2.6 announcement: a glass whale, marked-up and edited interiors, an owl sketch and animation posters. These are official reference examples, not images generated by this website.
Picsart / OpenAI
fal / OpenAI
fal / OpenAI
OpenAI
OpenAI · animation poster
OpenArt / OpenAI · animation posterPrompt guide
Write a structured brief and refine it with focused follow-ups. These prompt principles work across the GPT Image family; check each model’s supported features.
Prompt 01
Start with the subject and action, then add setting, camera distance, perspective, lighting, mood, and where key elements belong. If copy will sit above the subject, reserve that space in the brief.
Prompt 02
Put exact wording in quotation marks, specify where it belongs and describe the visual hierarchy. Always check important copy character by character.
Prompt 03
Name the element to change, the intended result and the details to preserve. Smaller, focused requests make it easier to judge whether a revision solves the problem.
Model comparison
OpenAI Image 2, OpenAI Image 2.5 and OpenAI Image 2.6 are common search terms for the GPT Image family. Compare official model capabilities separately from the models enabled in this studio.
01
The 2.6 release introduces further editing improvements, markup-guided changes, GIF workflows and API SVG output. Read the official release and model documentation for the exact supported surfaces.
02
GPT Image 2.5 introduced Flare and Sunburst choices with an emphasis on image quality, reference fidelity and editing. Its model guide remains available on Open AI Image.
03
Choose GPT Image 2 in the model picker for text-to-image generation or reference-led editing.
FAQ
Release details, available models, official examples and practical image workflows.
Related models
Explore GPT Image 2.5 and GPT Image 2, or compare other image models in the same workspace.
GPT Image 2.6 · Official release November 4, 2026
Explore GPT Image 2.6, then create with the image and video models currently available on Open AI Image.
Start Now