GPT Image 2.6 Sunburst

Not yet independently verified. Not yet independently verified. Release date, prices and capabilities rest on OpenAI's own changelog and model page; we found no dated independent report or test yet. We will update this when it can be confirmed, and remove this note.

GPT Image 2.6 Sunburst is the detailed-work model of the pair OpenAI released on 4 November 2026, alongside the faster GPT Image 2.6 Flare. It takes text and images as input and returns images, and OpenAI positions it for detailed compositions and editing workflows. GPT Image 2.5 Sunburst and Flare, from September, remain listed.

Complexity, markup and new output formats

OpenAI's changelog lists four additions with the 2.6 models. An adjustable complexity setting trades detail against speed. Markup-guided editing lets a caller mark the region of an image to change, and OpenAI says it also works with supported GPT Image 2.5 models. The model can return SVG vector output and animated GIFs as well as raster images, which matters for teams producing icons, diagrams or short animations that would otherwise need a separate tool.

Where it runs and what it costs

The model is served only through the Images API, on the image generation and image edit endpoints; the model page marks Chat Completions, Responses, Batch and fine-tuning as not supported. Pricing is per token and is the same for Sunburst and Flare: text input at $5 per million tokens ($1.25 cached), image input at $8 per million ($2 cached) and image output at $30 per million. The page states no context window.

Key facts

Developer: OpenAI Released: 4 November 2026 Model ID: gpt-image-2.6-sunburst Input: text, image Output: image (including SVG and animated GIF) Endpoints: images/generations, images/edits Price: $5 per million text input tokens, $8 per million image input tokens, $30 per million image output tokens