> For the complete documentation index, see [llms.txt](https://docs.pletor.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.pletor.ai/models/image-models/gpt-image-2.6.md).

# GPT Image 2.6

OpenAI's latest image model, now with animated GIF and SVG output. Two variants, Flare and Sunburst.

## Overview

The GPT Image model to reach for when the deliverable isn't a still image, or when a chain of edits has to hold. It keeps what [GPT Image 2.5](https://docs.pletor.ai/models/image-models/gpt-image-2.5.md) does well (text, layout, brand accuracy, the same quality scale) and adds animated GIFs, SVG vectors from a prompt, and a Reasoning setting that plans dense text and complex layouts before rendering. Edits stay closer to the approved look round after round.

| Variant              | Flare                                                     | Sunburst                                                              |
| -------------------- | --------------------------------------------------------- | --------------------------------------------------------------------- |
| Best for             | Fast iteration: social creative, drafts, GIFs, most edits | Detailed work: campaign finals, packaging, intricate product shots    |
| Speed                | Fast                                                      | Slower, trading time for fine detail                                  |
| Consistency in edits | Strong                                                    | Tightest: characters, products and brand details hold edit after edit |
| Output formats       | Image, GIF, SVG                                           | Image, GIF, SVG                                                       |

Start with Flare; move to Sunburst when a Flare edit drifts or the asset is a final. Sunburst is the default for markup edits.

### What's new in 2.6

* **Animated GIFs**: use the dedicated node to generate GIFs. Connect still images to guide the animation.
* **SVG vectors**: set **Output Format** to SVG for logos, icons and illustrations with editable paths. From a prompt only: once an image is connected, the choice is Image or GIF.
* **Reasoning**: Low, Medium or High (default). Higher helps dense text and complex layouts but is slower; lower suits focused edits.
* **Markup-guided edits**: circle, point at or annotate the area to change, then describe the change.

### Settings

* **Quality**: Low, Medium, High (default), Extra high, Max.
* **Resolution**: 1K to 4K (2K by default); aspect ratio Auto, or 21:9 to 9:21.
* **References**: up to 16 images.
* **GIF frame size**: frames share one image budget, so fewer frames leave each more room.

***

## How to use effectively

* **State the invariant** in edits: "Keep the product, its position, the lighting and the background unchanged."
* **Lower Reasoning for small edits**: High re-plans the whole image and can change things you didn't ask for.
* **Describe motion and loop for GIFs**: "returns to its starting pose" gives a clean loop.
* **Keep SVG for graphic work**: logos, icons, flat illustrations.
* **Pick the variant per step**: Flare for drafts, Sunburst for the final pass.
* **Be explicit with copy**: exact text in quotes, with placement.

**Example prompts**

* **GIF product loop (Flare)**: "The perfume bottle from the reference turns slowly through a full rotation on a pale pink pedestal, soft studio light. The loop ends where it starts. Keep the label legible in every frame."
* **SVG icon (Flare)**: "A flat vector icon of a delivery van, two colors: #1B3A5C and #F2B134. Rounded corners, no text."
* **Localization (Reasoning: Low)**: "Replace the headline 'Soft hands, every day.' with 'Des mains douces, chaque jour.' Same font, weight, color and position."
