Independent testing confirms the capabilities of GPT Image 2.5 Sunburst and Flare from OpenAI, including editing with a reference image
Commentator Simon Willison tested the new image models from OpenAI, GPT Image 2.5 Sunburst and Flare, recommends Sunburst for precise edits and Flare for fast generation, and also verified editing using a reference image.
Independent LLM commentator Simon Willison tested both newly released image generation models from OpenAI, GPT Image 2.5 Sunburst and GPT Image 2.5 Flare. According to his assessment, Sunburst is a suitable choice for workflows where editing precision is key, while Flare is intended for fast, high-quality routine image generation. Willison also extended his command-line tool openai_image.py to support one or more reference images and tested this feature in practice by editing an existing photograph – he used Sunburst to add a new element to the scene based on a text prompt.
OpenAI released GPT Image 2.5 Sunburst and Flare on 8 September 2026 for image generation and editing through Image API and the image generation tool in Responses API. According to the company, both models support the new quality levels xhigh and max and use the same token rates as GPT Image 2.
Why it matters
Independent testing of two new image generation models gives developers specific guidance on when to choose the more precise but probably slower Sunburst variant and when to choose the faster Flare variant, without having to conduct their own comparative tests. Verified support for editing using a reference image also shows that image-to-image edits work in readily available third-party tools, expanding the possibilities for practical deployment in photo editing applications.
Release card
GPT Image 2.5 Sunburst
OpenAI
- Inputs
- image generation and editing, including editing with one or more reference images
- Availability
- Image API (v1/images) and the image generation tool in Responses API
- image edits where precision is key
- editing an existing image using one or more reference images
- for fast, routine image generation, the source recommends Flare instead
According to the source, the model is positioned relative to Flare, which was released at the same time: Sunburst is intended for edits where precision is key, while Flare is intended for fast, high-quality routine image generation.
The card summarizes information from the article and any dated corrections, with a link to the original source. It is not our assessment of the model. It does not yet have a dedicated editorial profile. Model selection and other announcements →
What was added since the original report
Verified updates
-
The model supports image-to-image generation with a reference image for editing; Simon Willison (an authoritative LLM commentator) tested both models and recommends Sunburst for precise edits, Flare for speed; Practical verification: the CLI tool integrates reference image support for image generation
- The model supports image-to-image generation with a reference image for editing
- Simon Willison (an authoritative LLM commentator) tested both models and recommends Sunburst for precise edits, Flare for speed
- Practical verification: the CLI tool integrates reference image support for image generation
Two audiences, two different impacts
What this means
For individuals
Developers and content creators working with image generation have a specific, practically tested recommendation on which of the two new models to choose based on the task type – precise editing or fast generation.
For a business
Companies building products on Image API from OpenAI have an independently verified choice between two models with the same token rates as GPT Image 2, making architecture decisions easier without the need for their own testing.
DevelopmentCheck the original
Event sources
independently confirmed · 2 publishers, 1 independent. We count feeds from the same owner only once.