GPT Image 2.6 Sunburst
GPT Image 2.6 Sunburst is the November 4, 2026 update to GPT Image 2.5 Sunburst. It generates images from text and edits existing images, with transparent backgrounds and custom dimensions. The model documentation positions Sunburst for intricate detail and premium visual work, with longer generation times.
Anvisha Pai, Co-founder & CEO, Voyager
OpenAI · released
›What we asked for
A seamless looping animation of a small orange paper-cut sun slowly spinning above three gently rolling blue ocean waves. Nine distinct frames, fixed camera, clean cream background, bold simple shapes, no text.
openai/gpt-image-2.6/sunburst/text-to-image
image_size: {"width":832,"height":832}, quality: medium, output_format: gif, frame_count: 9, frame_duration_ms: 120, reasoning_effort: low, num_images: 1
returned 832 by 832 pixels
At a glance
- Same token prices
- The 2.6 update keeps the 2.5 rates. Quality, dimensions, reasoning and reference images affect the final cost.
- Quality and reasoning
- Quality defaults to high. Reasoning effort is a separate setting, from low through max.
- Transparent backgrounds
- Choose transparent, opaque or automatic backgrounds.
- Image editing
- Supply up to 16 reference images. Use a mask or a separate annotation image to guide an edit.
Other GPT Image models: GPT Image 2.6 Flare, GPT Image 2.5 Flare, GPT Image 2.5 Sunburst, GPT Image 2. Compare all 5.
Anvisha's take
We would start new GPT Image work with 2.6 and keep 2.5 for projects already tuned to it. The listed token prices are unchanged, so moving versions does not itself raise the rate. We generated the nine-frame GIF above in one attempt. We have not run our fixed prompt set on 2.6 yet: we cannot claim a measured improvement in speed, text accuracy or edit fidelity, and the 2.5 results remain on their own pages.
Recorded benchmark samples coming later
We have verified GPT Image 2.6 Sunburst’s documentation and pricing, but have not run our fixed prompt set yet.The example above is a separate run, not part of the common benchmark.
GPT Image 2.6 Sunburst pricing
| Where | Unit | Price | Source |
|---|---|---|---|
| API list price | Per million text input tokens | $5.00 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | Per million cached text input tokens | $1.25 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | Per million text output tokens | $10.00 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | Per million image input tokens | $8.00 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | Per million cached image input tokens | $2.00 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | Per million image output tokens | $30.00 | GPT Image 2.6 Sunburst generation API reference, read |
| API list price | 1024 by 1024, high quality Same size and quality estimate as 2.5. Prompt and reference tokens add cost; the request total is rounded up to $0.0001. This is an estimate, not a flat fee. | $0.053 | GPT Image 2.6 Sunburst generation API reference, read |
Is GPT Image 2.6 Sunburst free?
API generation is billed by token usage. This release reference does not establish a free allowance for GPT Image 2.6.
- •Voyager offers GPT Image generation on paid plans using credits.
- •ChatGPT access and limits are separate from these API endpoints.
Where to use GPT Image 2.6 Sunburst
- Voyager · Generate and edit images using Voyager credits on paid plans.
- Generation API reference · Documented inputs, outputs and token pricing.
- Edit API reference · Reference images, masks and annotation inputs.
Voyager
Run GPT Image 2.6 Sunburst inside Voyager
Voyager is an agent for creative work that runs every model in this reference, GPT Image 2.6 Sunburst included. Brief it, and it picks the model, makes the thing and hands back something you can edit. Create an account to get started.
What we noticed running GPT Image 2.6 Sunburst
Not written up yet.
GPT Image 2.6 Sunburst limits and API parameters
| Limit | Value |
|---|---|
| Output size | Dimensions in multiples of 16, longest edge 3840 pixels, aspect ratio at most 3:1, and 655,360 to 8,294,400 total pixels. |
| Reference images | Up to 16 per edit request. |
›Every setting, for developers
Generation API
GPT Image 2.6 Sunburst generation API reference, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| prompt | string | Any text Required. | |
| image_size | enum or object | Preset name, auto, or {width, height} Explicit dimensions must be multiples of 16; longest edge 3840 pixels, aspect ratio at most 3:1, total pixels 655,360 to 8,294,400. | landscape_4_3 |
| quality | enum | auto, low, medium, high | high |
| reasoning_effort | enum | low, medium, high, xhigh, max Higher settings increase detail, latency and token usage according to the API reference. | high |
| background | enum | auto, transparent, opaque | auto |
| num_images | integer | 1 to 10 SVG output requires one image. GIF frames share the request pixel budget. | 1 |
| output_format | enum | jpeg, png, webp, gif, svg | png |
| output_compression | integer | 0 to 100 JPEG and WebP only. | |
| frame_count | integer | 2 to 64 GIF only; each frame uses image_size. | 9 |
| frame_duration_ms | integer | 20 to 2000 GIF only; rounded to 10 milliseconds. | 200 |
Edit API
GPT Image 2.6 Sunburst edit API reference, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| image_urls | list of URLs | Up to 16 Required with prompt on the edit endpoint. | |
| mask_url | URL | A mask image White marks regions to edit; black marks regions to keep. | |
| markup_image_url | URL | A transparent PNG containing annotations Must match the first reference image dimensions. Cannot be combined with mask_url. | |
| image_size | enum or object | Preset name, auto, or {width, height} Auto infers dimensions from the input images. | auto |
| output_format | enum | jpeg, png, webp, gif The edit endpoint does not list SVG output. | png |
What GPT Image 2.6 Sunburst will not draw
Safety checking is enabled by default. The documented option for relaxed screening is restricted to authorized callers; it is not a guarantee that a prompt will be accepted.
Alternatives to GPT Image 2.6 Sunburst
Frequently asked questions
Does 2.6 cost more than 2.5?
The listed token rates are identical. At the same size and quality the per-image estimates are unchanged, including about $0.053 for 1024 by 1024 at high quality. Inputs and reasoning can change the final bill.
Have you tested GPT Image 2.6?
Only lightly. We have verified the API documentation and pricing and generated the nine-frame GIF on this page, but we have not run our fixed prompt set on 2.6, so there are no benchmark results to show yet.
Are xhigh and max quality options?
In the 2.6 API reference, quality accepts auto, low, medium and high. The separate reasoning_effort setting accepts low, medium, high, xhigh and max. Both default to high.
How this page is made
Prices, limits and settings are read from the linked vendor and host pages on the dates shown. Every picture was generated by us through fal with the request shown under it, and the original files are kept unedited. A generation that failed is shown as a failure. Model and vendor names are used to identify the products; no vendor imagery appears on this page.
