GPT Image 2.5 Sunburst

• - 🆕 Editing variant of OpenAI's GPT Image 2.5 Sunburst, released September 2026.
• - 🔧 Built for precision editing: masked inpainting and outpainting workflows.
• - 🎯 OpenAI recommends Sunburst where editing precision matters most.
• - 👁️ Accepts both text and image inputs; edits existing images.
• - 📏 Quality tiers include low, medium, high, xhigh, max and auto.
• - ⚡ Trades longer generation time for extra precision versus Flare.
• - 🏢 Selectable in the Image API or as the Responses API image tool.

OpenAI

OpenAI is an American artificial intelligence research organization headquartered in San Francisco, structured as both a for-profit public benefit corporation and a nonprofit foundation. The lab developed the GPT family of large language models, the DALL-E…

GPT Image 2.5 Sunburst (edit) is the image-editing entry point for OpenAI's GPT Image 2.5 generation, released alongside GPT Image 2.5 Sunburst and the faster GPT Image 2.5 Flare pair, including GPT Image 2.5 Flare in edit mode. OpenAI describes Sunburst as generating and editing images from text and image inputs, aimed at workflows where editing precision matters most, with selectable quality levels from low up through xhigh and max.

Compared with the previous family generation, GPT Image 2 and its editing counterpart GPT Image 2, OpenAI reports improvements in detail, style adherence and control over edits across the 2.5 models. That added edit control is what makes this variant a practical fit for text-aware inpainting and typography-heavy assets, where surrounding composition should stay intact.

Within the 2.5 tier, the trade-off is explicit from OpenAI: Flare is the recommended default for most applications and delivers higher-quality images than GPT Image 2 at 50% lower latency, while Sunburst spends longer generating in exchange for an extra level of precision on detailed creative work.

Access is via the Image API by naming the Sunburst model directly, or as the image-generation tool inside the Responses API for conversational, iterative editing sessions; a parameter controls whether a request generates a new image or edits one already in context. Masked inpainting and outpainting are supported through the edits endpoint. Token rates match GPT Image 2, though OpenAI notes the older calculator does not estimate 2.5 token consumption. Earlier family members such as GPT Image 1.5 and GPT Image 1.5 remain available.