Image generation

ChatGPT Images 2.6

by OpenAI

OpenAI's image model, released November 2026.
Animated GIFs, SVG export, faster drafts.

ChatGPT Images 2.6

Key features

Sourdough loaf on an oak bench, steam risingStart creating

Technical specifications

Nov 4, 2026

In ChatGPT and Codex, on every plan.

Two

Flare for speed, Sunburst for detail and steadier edits.

Up to 4K

Up to 3840px on the long edge, 1:3 to 3:1.

Three levels

Low, medium or high. High is the default.

Use cases

Looping mascots

A brand character that kneads, waves or bounces on repeat, for social posts, stickers and app moments.

Product loops

One still with steam, light or flour dust moving through it: a scroll-stopping loop without a video shoot.

A rust and green riso poster reading AUTUMN BAKE and EMBER & OAT taped inside a bakery window

Posters and layouts

OpenAI says Flare lays out posters and other professional visuals better. Quote every word you want printed.

The same pastry box three times: plain, then tied with a rust ribbon, then with a round brand sticker

Edits that hold

Change one thing per round and keep the rest. Sunburst is the model OpenAI built for steadier edits.

A character sheet of a dough baker mascot with a chef hat and rust neckerchief from front, side and back

Reusable characters

Build a reference sheet once, then reuse it in every new scene so the face and outfit stay the same.

The Ember and Oat bakery logo beside six rust line icons of a loaf, croissant, wheat, whisk, oven and cup

Logos and icons as SVG

Flat marks and icon sets that need to scale. On Morphic, make them as SVG with Ideogram Vector.

Prompt examples

A cream riso poster reading AUTUMN BAKE in rust serif above a scored loaf

Riso poster

Riso poster, AUTUMN BAKE in rust serif, a scored loaf on cream paper, no other text

Edit prompt
A sourdough loaf on a floured oak bench with steam rising in dawn window light

Bakery still

Sourdough loaf on a floured oak bench, dawn window light, rising steam, film look

Edit prompt
Six frames of a papercraft koi swimming in a circle, arranged in a three by two grid

Loop frames

Sprite sheet, 6 equal frames in a 3x2 grid: a papercraft koi swimming a slow circle

Edit prompt

Overview

ChatGPT Images 2.6, also written GPT Image 2.6, is the image model OpenAI released on November 4, 2026. It runs in ChatGPT on every plan and in Codex, as two models: GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst.

The headline change is motion: it can return a short looping GIF from a prompt or from an image you upload. Around that, OpenAI added generation it measures as 50% faster than before, a complexity setting for how much the model plans a scene, SVG export, image markup that shows where an edit goes, and saved reference images in ChatGPT that you call back by name.

GPT-Image-2.6 Flare or Sunburst

Flare is the fast model, able to return an image in about six seconds at low quality, with better layouts for posters and product shots. Sunburst takes longer for more detail and holds characters and products steadier across a chain of edits. Start with Sunburst when the image has to be exact, then test whether Flare clears the same bar faster.

ChatGPT Images 2.6 vs 2.5

The size limits, the five quality tiers and the prices carry over from 2.5. The gain is in what the model can return and how much control you get. On image quality alone, 2.6 has not yet pulled ahead: on Arena's preliminary blind-vote scores, read on November 5, 2026, both 2.6 models sat just behind their 2.5 versions.

ChatGPT Images 2.6 and Morphic

ChatGPT Images 2.6 is not in the Morphic catalog yet. GPT Image 2.5 Flare and Sunburst are, on the paid plans, for generating and for editing with up to 10 reference images. The examples on this page were made on Morphic with the closest live model to each 2.6 feature: GPT Image 2.5 for stills and edits, Ideogram Vector for the vector art, and MiniMax H3 Max Turbo for the loops. For prompting tips, read the ChatGPT Images 2.6 guide, copy a starting point from the ChatGPT Images 2.6 prompt library, or open the GPT Image 2.6 AI image editor.

Simple pricing

Get started for free today, with the option to upgrade or cancel anytime.

Basic

$19/ month
billed as $228 per year

2400 monthly credits

1 user only

All models

Workflows

Standard

$24/ month
billed as $288 per year

3625 monthly credits

1 user only

All models

Workflows

Pro

$45/ month
billed as $540 per year

6350 shared monthly credits

1 user

+ up to 4 more at extra cost

All models

Workflows

Max

$170/ month
billed as $2040 per year

24650 shared monthly credits

1 user

+ up to 9 more at extra cost

All models

Workflows

Enterprise

For higher limits

Custom

pricing and billing terms

High-volume credits
Custom seat limits
All models
Workflows
Pricing Gradient

Free

For playing around

$0

forever free

Up to 20 credits
1 user only
Limited models
Workflows
Compare plan details

FAQs

What is ChatGPT Images 2.6?
ChatGPT Images 2.6, also written GPT Image 2.6, is the image model OpenAI released on November 4, 2026. It adds animated GIFs, SVG export, a complexity setting and saved reference images in ChatGPT, and OpenAI says it generates 50% faster than before. It comes as two models, GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst.
How do I make a GIF with ChatGPT Images 2.6?
In ChatGPT, ask for a looping GIF, or upload a still and ask to animate it, then adjust the speed in the chat. On Morphic, ask Copilot for a looping clip in one message: describe the still, then the motion, and say it should end where it starts. Copilot makes the image and animates it, and the clip lands on your Canvas.
What is the difference between GPT-Image-2.6 Flare and Sunburst?
GPT-Image-2.6 Flare is the smaller, faster model; OpenAI says it can produce an image in about six seconds at low quality and handles posters and product layouts better than before. GPT-Image-2.6 Sunburst is the base model, built for more detail and more consistency across a chain of edits at medium or high quality. Start with Flare when speed matters and with Sunburst when the image has to be exact.
What does the complexity setting do?
Complexity controls how much the model plans a scene before it renders, separately from quality, which controls detail. It has three levels: low, medium and high, with high as the default. OpenAI suggests low for quick, focused edits and medium or high for scenes with many elements or constraints, and notes that higher settings can add time. On Morphic, get the same effect by writing the plan into the prompt: list every label once and say where text is not allowed.
What is new in ChatGPT Images 2.6 compared to 2.5?
Animated GIF output, SVG output, the complexity setting, image markup for edits, and saved reference images in ChatGPT. The two-model setup, the five quality tiers from low to max, and the size rules (up to a 3840px edge, aspect ratios from 1:3 to 3:1) carry over from 2.5.
Is ChatGPT Images 2.6 better than ChatGPT Images 2.5?
It does more, but it is not yet shown to look better. On Arena's blind-vote image leaderboards read on November 5, 2026, both 2.6 models sat slightly below their 2.5 counterparts, on preliminary scores from a few thousand votes. The reasons to pick 2.6 today are its new outputs and controls: GIFs, SVG, complexity and faster drafts. Compare both on your own prompts.
Can ChatGPT Images 2.6 export SVG files?
Yes. ChatGPT Images 2.6 can return a logo or icon as an SVG vector file, drawn from a reference image it renders first. On Morphic today, Ideogram Vector makes SVG files for logos, icons and flat illustrations; keep the design to a few solid colors so the paths stay clean.
Are ChatGPT Images 2.6 images watermarked?
They carry provenance data rather than a visible watermark. OpenAI attaches C2PA metadata and an invisible SynthID watermark across ChatGPT and Codex, so a checker can later identify the image as AI-generated.

More about ChatGPT Images 2.6

How to use ChatGPT Images 2.6: GIFs, edits, and the two models

ChatGPT Images 2.6 (GPT Image 2.6) on Morphic: looping animations with Copilot, Flare vs Sunburst, busy scenes, one-change edits, consistent characters, and vector logos.

Other models

SAM 3.1

Meta

Meta's Segment Anything model. Track any subject as a mask that follows it.

Seed Audio 1.5

ByteDance

ByteDance's full-scene audio model. Dialogue, music and effects, as separate tracks.

FLUX Video Upscale

Black Forest Labs

Black Forest Labs' video upscaler. Up to 4K, with fine detail rebuilt and the sound kept.

Ideogram Vector

Ideogram

Ideogram's model for vector art. Logos, icons and illustrations as SVG files.