- Precise edits that keep products and characters
Built for
- Up to 10, with optional mask
Input images
- Low, medium, high, xhigh, max
Quality levels
- Low, medium, high (default high)
Planning effort
- 3840px edge, 1:3 to 3:1
Max size
About GPT Image 2.6 Sunburst
GPT Image 2.6 Sunburst is the detailed model in OpenAI's GPT Image 2.6 pair, built for precise edits and for keeping products, characters, and brand details faithful to the images you supply. That is the hard part of business creative: changing one element of an approved image without disturbing the rest, placing a real product into a new scene, or updating a layout that already carries brand text.
In Masonry, Sunburst 2.6 keeps the controls of GPT Image 2.5: up to 10 input images per request, an optional mask, one to four images at a time, low to max quality, transparent backgrounds, and any size OpenAI allows up to 3840 pixels on the long edge. It adds one control, complexity, which sets how much the model plans before it renders, separately from rendering quality. Its sibling, GPT Image 2.6 Flare, is the fast model for interactive generation.
Why teams choose GPT Image 2.6 Sunburst
Choose GPT Image 2.6 Sunburst when an edit has to keep an uploaded product, character, or brand detail faithful, and when you want separate control over how much the model plans and how finely it renders. Use Flare for fast exploration and Sunburst for the finished deliverable. We have not run Sunburst 2.6 through Masonry's own tests yet, so compare it with GPT Image 2.5 Sunburst on your own source images before switching a production workflow.
What GPT Image 2.6 Sunburst can do
The capabilities that set GPT Image 2.6 Sunburst apart and earn its place in a brief
Built for Precise Edits
Sunburst 2.6 is the GPT Image 2.6 model for edits that must keep products, characters, and brand details intact. Use it to swap a background, change a color, replace a prop, or update copy on an image you have already approved.
Planning Effort, Separate From Quality
The new complexity setting controls how much the model plans before it renders, independent of quality. Low suits a focused edit; high, the default, suits complex layouts with several elements and text.
Up to 10 Input Images
Add up to 10 images to a request to edit or reference. Supplying any image switches the request to edit mode, so a product shot, a logo, and a style reference can all inform one result.
Masks for Targeted Changes
Add an optional mask to mark the area to change, and describe the edit clearly in the prompt as well, since the mask guides the change rather than acting as a hard boundary.
Five Quality Levels, Up to Max
Choose low, medium, high, xhigh, or max. Masonry defaults to low, so explore cheaply and raise quality for the final render.
Any Size Up to 3840 Pixels
Pick from 28 presets, from 1024x1024 to 3840x2160 and including 2880x2880 squares, or set a custom size. Edges are multiples of 16, the long edge tops out at 3840 pixels, and ratios run from 1:3 to 3:1.
Where teams reach for GPT Image 2.6 Sunburst
- Product-in-scene photographySupply a real product photo and describe the scene, so the product appears in a lifestyle or studio setting without a reshoot. Check labels and fine detail against the source before publishing.
- Targeted edits to approved imagesChange the background, color, prop, or season of an approved hero image while keeping everything else as it is. Try low complexity for a single, focused change.
- Character and brand consistencyUse a character sheet, logo, or palette as input images to keep a recurring character or brand look consistent across a set of ads.
- Layouts with textPosters, ad frames, and infographics where headlines and labels must be spelled exactly. Keep complexity high for dense layouts and quote every string in the prompt.
- Transparent cut-outsGenerate isolated products and objects as transparent PNGs for marketplaces, decks, and templates.
- Finishing a pick from FlareExplore options quickly on GPT Image 2.6 Flare, then continue on Sunburst with the chosen image as input for the edits and the final render.
Frequently asked questions
What teams need to know about creating with GPT Image 2.6 Sunburst in Masonry
What is GPT Image 2.6 Sunburst?
GPT Image 2.6 Sunburst is OpenAI's detailed GPT Image 2.6 model for precise edits and for preserving products, characters, and brand details. It generates images from text and edits images you supply, including masked edits.
What is new in GPT Image 2.6 compared with 2.5?
In Masonry, the 2.6 models keep the 2.5 controls (up to 10 input images, a mask, one to four outputs, low to max quality, transparent backgrounds, and sizes up to 3840 pixels) and add complexity, a setting for how much the model plans before it renders, separate from rendering quality.
What does the complexity setting do?
Complexity sets planning effort, independent of quality. Low suits focused edits, such as changing one object or the background. High, the default, suits complex layouts with several elements or lines of text. Medium sits between them.
Should I use GPT Image 2.6 Flare or Sunburst?
Use Flare, the fast model, for interactive generation and quick rounds of options. Use Sunburst, the detailed model, when an edit has to keep a product, character, or brand detail faithful, or for the final render.
Can GPT Image 2.6 Sunburst edit an existing image?
Yes. Add one or more input images, up to 10, and the request switches to edit mode. Describe what should change and what should stay the same, and add a mask to mark the area to change.
What sizes and aspect ratios does GPT Image 2.6 Sunburst support?
Masonry offers 28 presets from 1024x1024 to 3840x2160, including 2880x2880 squares, and accepts custom sizes where both edges are multiples of 16, the long edge is at most 3840 pixels, and the ratio is between 1:3 and 3:1.
Can GPT Image 2.6 Sunburst make transparent PNGs?
Yes. Set the background to transparent and the output is a PNG with no backdrop. The other options are auto and opaque.
How many images can GPT Image 2.6 Sunburst generate at once?
One to four per request in Masonry.
Has Masonry tested GPT Image 2.6 Sunburst?
Not yet. This page lists the route's controls as checked on November 5, 2026. Our first-hand results so far cover GPT Image 2 and the 2.5 models, which rendered every line of our seven-line label test exactly. Compare 2.6 with 2.5 on your own source images before switching a production workflow.
How does my team use GPT Image 2.6 Sunburst in Masonry?
Open a Masonry canvas, pick GPT Image 2.6 Sunburst 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 2.6 Sunburst with other models in one workspace.
Is GPT Image 2.6 Sunburst free to try?
New Masonry accounts get 50 free credits to try GPT Image 2.5 Flare. GPT Image 2.6 Sunburst is included in every paid plan, and the editor shows the exact credit cost before you generate.
How do I write good prompts for GPT Image 2.6 Sunburst?
For an edit, say what should change and what must stay the same, in that order, and name the input image each instruction refers to when you supply several. Set complexity to low for one focused change and leave it at high for layouts with several elements or lines of text. Put every on-image string in quotation marks with its position, explore at low quality, and raise quality only for the final render.
Who makes GPT Image 2.6 Sunburst?
GPT Image 2.6 Sunburst 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.
Start creating with GPT Image 2.6 Sunburst
Generate, edit, and compare across 50+ models in one workspace.
Guides for GPT Image 2.6 Sunburst
Prompt walkthroughs and examples from the Masonry blog
Explore more AI models
Compare GPT Image 2.6 Sunburst with other models teams run in Masonry