OpenAI's GPT Image 2.6 comes as Sunburst and Flare. What the new complexity setting does, which one to pick, how to use it on CRAISEE and what it costs.
Frequently asked questions
What is the difference between GPT Image 2.6 Sunburst and Flare?
GPT Image 2.6 Sunburst is designed for detailed compositions and precise edits — ideal for posters, packaging, infographics, and UI mockups. GPT Image 2.6 Flare is optimized for fast iteration, making it the better choice for drafts, quick concepts, and creative variations. Both support the same quality levels, sizes, and the new complexity setting.
What does the complexity setting do in GPT Image 2.6?
The complexity setting controls how much the model plans an image before rendering it. High complexity allows more planning, improving results for dense text and complex layouts like menus or infographics. Low complexity is faster and suits simple scenes. OpenAI defaults to high when no complexity level is specified, and it costs nothing extra.
Does complexity cost more credits on CRAISEE?
No. OpenAI charges nothing extra for the complexity setting, so using high, medium, or low complexity in GPT Image 2.6 does not change the credit price on CRAISEE. The only trade-off is that higher complexity can make an image take longer to generate.
When should I use GPT Image 2.6 Sunburst vs Flare?
Use Sunburst when precision matters — detailed compositions, product labels, infographics, or edits where unchanged areas must stay intact. Use Flare when speed matters — drafts, creative exploration, and rapid iteration. OpenAI explicitly describes Flare as the model for fast iteration.
What is the maximum image size supported by GPT Image 2.6?
Both GPT Image 2.6 Sunburst and Flare support a maximum image size of 3840×2160 pixels. Sizes above 2560×1440 are currently experimental. This applies to both text-to-image generation and editing with reference images on CRAISEE.
Can GPT Image 2.6 edit images with reference photos on CRAISEE?
Yes. Both GPT Image 2.6 Sunburst and Flare support editing with reference images on CRAISEE, accepting up to 16 reference images per request. This capability carries over from GPT Image 2.5 and is available across all quality levels.
Does GPT Image 2.6 support transparent backgrounds?
Yes. Both Sunburst and Flare versions of GPT Image 2.6 support transparent backgrounds, exported in PNG or WebP format. This feature is inherited from GPT Image 2.5 and is available on CRAISEE at all quality levels.
Why does higher complexity cause unwanted changes in edits?
OpenAI notes that higher complexity levels allow the model to plan more broadly before rendering, which can cause it to reinterpret or alter areas of an image you didn't intend to change. If an edit drifts unexpectedly, switching to a lower complexity level reduces the model's planning scope and helps preserve the original image more faithfully.
