OpenAI Images 2.6: GIFs and 50% Faster Images

Updated: November 20265 min readBy the CreatorSprout team

Launch coverage based on OpenAI’s November 4, 2026 announcement and its published API pricing, as reported by multiple outlets, we have not tested Images 2.6 hands-on. Speed and quality claims below are OpenAI’s, not independent test results. No affiliate links in this article.

Images 2.6 is not a new product. It is the image engine inside ChatGPT getting faster, more editable, and, for the first time, able to animate. That matters for creators because the most useful part of the update is not a benchmark number. It is that a still image you already made can now become a moving one without leaving the chat, exporting to another tool, or paying for a separate animation app. The GIF is the story; the speed and the two API models are what make it practical at volume.

What was announced

OpenAI released Images 2.6 on November 4, 2026, across ChatGPT, Codex and the API, replacing Images 2.5, which had shipped in September. The release has four parts: faster generation, GIF output, two distinct API models, and a set of finer editing controls, SVG export, markup-guided edits, and a complexity setting that adjusts how much planning the model puts into a generation. OpenAI says people now generate more than 4 billion images per week across ChatGPT and the API, which explains why the company is splitting the model in two: at that volume, a fast draft model and a careful final model serve different jobs better than one compromise model.

The rollout is global and covers every subscription tier, including the free plan, on web, desktop and mobile. The only staged part is GIF creation itself, which OpenAI is rolling out gradually, if you do not see it yet, that is the rollout, not a plan restriction.

GIF creation: the headline feature

Until now, ChatGPT could make still images and you animated them somewhere else. Images 2.6 closes that gap:

Three ways in. Describe a motion in plain language, upload a hand-drawn sketch, or start from an existing image, including a character, mascot or product shot, and the model returns a short looping animation.
Frame-by-frame editing. You can edit individual frames and adjust animation speed, rather than re-rolling the whole GIF when one frame looks wrong.
Built for sharing. The output is a looping GIF you can share directly, the format social platforms, group chats and product pages already accept, with no video player or conversion step.
Sprite-sheet to animation. The model can also assemble a multi-frame sprite sheet into a smooth animation, which is handy for creators who already work with frame grids.

For creators, the practical use cases are obvious: a mascot that waves in a community post, a product that spins in a listing, a reaction GIF in your own style instead of a stock one, a sketch that becomes a moving storyboard panel. None of this replaces a video generator for longer clips, these are short loops, not scenes, but for the huge middle ground of “I want this image to move a little,” it removes a whole tool from the workflow.

Flare vs Sunburst: two API models

Developers do not get one GPT-Image-2.6 model. They get two, with the same token pricing but different tuning:

GPT-Image-2.6 Flare, the fast one. Tuned for speed and everyday creation. OpenAI says it can return a low-quality image in as little as six seconds, and reports improved layout quality for posters and product shots. Use it for iteration: drafts, variations, and interactive workflows where waiting is the cost.
GPT-Image-2.6 Sunburst, the careful one. Tuned for image quality and consistency across repeated edits, at medium or high quality settings. Use it for the final version, detailed compositions, and multi-turn editing where the subject must stay the same from one edit to the next.

The split is the same idea as a fast draft mode and a slow final mode in any creative tool, except here it is two models with one price list, so the choice costs nothing extra. The API also lists SVG output and animated GIF output, so developers can generate vector graphics and animations programmatically, not only through the ChatGPT interface.

Speed and quality improvements

50% faster generation. OpenAI says Images 2.6 generates images 50% faster than Images 2.5. That is the company’s own comparison, not an independent benchmark, treat it as a directional claim until third-party testing exists.
Sharper rendering where creators notice it. Reported improvements cover lighting, skin tones, materials, and dense multi-column text layouts, menus, infographics, calendars and slides, plus better poster and product typography. Text-heavy images have long been the weak spot of image models, so this is the improvement to watch in real use.
More control over edits. Markup-guided edits let you mark up an image to direct where a change should happen, and the complexity setting lets you trade planning time against output detail. Both are aimed at the same frustration: re-rolling an entire image when only one corner needs fixing.
Consistency across edits. Sunburst is specifically tuned to keep a subject looking the same through repeated edit rounds, the difference between a usable character and one that drifts into a stranger by the third revision.

Pricing and availability

ChatGPT and Codex: included across all subscription tiers, including the free plan. OpenAI has announced no separate consumer price for Images 2.6, and the announcement does not specify new per-plan generation limits, so do not confuse “available on free” with “unlimited on free.”
API pricing, unchanged from Images 2.5, identical for Flare and Sunburst: text input $5.00 per million tokens ($1.25 cached), image input $8.00 per million tokens ($2.00 cached), image output $30.00 per million tokens. Pricing is token-based, not a flat price per picture, OpenAI points developers to its image-generation calculator for cost estimates.
Availability: live globally in ChatGPT and Codex on web, desktop and mobile. GIF creation is rolling out gradually and may not have reached your account yet.
Formats: PNG, JPEG, WebP, SVG and animated GIF output through the API.

What it means for creators

GIFs are now a ChatGPT-native format. If your content uses reaction GIFs, animated product shots or looping social posts, test whether Images 2.6 replaces a separate animation step before paying for one. The frame-by-frame editing is the feature that makes this practical rather than a gimmick.
Free-tier access matters. Because Images 2.6, including GIF creation as it rolls out, is available on the free plan, this is a capability any creator can try without a subscription decision. Compare it against the image tools you already pay for before renewing them.
Flare changes the iteration math for developers. Six-second drafts at the same token price as before mean more variations per dollar of patience, if not per dollar of tokens. If you build creator tools on the image API, Flare is the model to prototype with and Sunburst the one to ship finals from.
Text-heavy images are the test to run. Menus, infographics, calendars and slides are exactly the images creators actually need AI to get right, and the ones it has historically mangled. Run your own before/after on a real layout before trusting the improvement claims.
Don’t rebuild your workflow on day one. GIF creation is rolling out gradually, and no independent benchmarks exist yet. Try it on a low-stakes post first; keep your existing animation tool until you have seen a GIF survive your own quality bar.

The honest caveats

We haven’t tested it. Everything here comes from OpenAI’s announcement and its published API pricing as reported by multiple outlets. The 50% speed claim, the six-second Flare figure and the quality improvements are OpenAI’s claims, not results we measured.
No benchmark scores were published for 2.6. Unlike the model launches that ship with benchmark tables, this release comes with company performance claims and no independent scores to check them against yet.
GIF limits are unspecified. The coverage we reviewed does not state maximum GIF length, frame counts or resolution, and short looping GIFs are not a substitute for video. Expect limits to surface as creators test it.
Free does not mean unlimited. OpenAI has not published new per-plan generation limits for Images 2.6. Availability across all tiers is not the same as unlimited generation on the free plan.
The rollout is gradual. If GIF creation is missing from your account, that is the staged rollout OpenAI described, not a bug, and not a reason to pay for a higher tier to get it sooner.

Verdict

Images 2.6 is a genuinely useful update, and the GIF is the reason to care. A 50% speed gain is welcome but unverified, and two API models are a developer nicety, but native GIF creation inside ChatGPT, free-tier included, removes a real step from a real creator workflow: still image to moving image, no export, no second tool. The honest version of the excitement is the usual one: try it on a throwaway post, check the text rendering on your own layouts, and keep your existing animation workflow until the GIFs prove themselves. If they do, this is the image update that quietly saves creators a subscription.

FAQ

What is new in OpenAI Images 2.6?

Images 2.6, released November 4, 2026, generates images 50% faster than Images 2.5 according to OpenAI, adds looping GIF creation inside ChatGPT, and ships two API models, GPT-Image-2.6 Flare for speed and Sunburst for quality. It also adds SVG export, markup-guided edits and a complexity setting.

Can ChatGPT make GIFs now?

Yes, with Images 2.6 you can turn a text prompt, a sketch or an existing image into a short looping GIF inside ChatGPT, with frame-by-frame editing and speed control. GIF creation is rolling out gradually, so some accounts will not see it immediately.

What is the difference between Flare and Sunburst?

Both are GPT-Image-2.6 API models with identical token pricing. Flare is tuned for speed and everyday creation, OpenAI says it can return a low-quality image in as little as six seconds. Sunburst is tuned for image quality and consistency across repeated edits at medium or high quality.

How much does Images 2.6 cost?

In ChatGPT it is included across all tiers, including the free plan, with no separate price announced and no new per-plan generation limits published. Through the API, pricing is unchanged from Images 2.5: $5 per million text-input tokens ($1.25 cached), $8 per million image-input tokens ($2 cached) and $30 per million image-output tokens.

Is Images 2.6 available to everyone?

OpenAI says Images 2.6 is available globally in ChatGPT and Codex across all subscription tiers on web, desktop and mobile. The one exception is GIF creation, which is rolling out gradually, so it may not appear in your account yet.