OpenAI Images 2.6 Is Out: Features, Pricing and Access

What is OpenAI Images 2.6?

OpenAI Images 2.6 is an image generation update from OpenAI, announced November 4, 2026, that adds GIF creation, reusable reference images and faster generation to ChatGPT, Codex and the API. OpenAI says it generates images 50% faster than before.

OpenAI's announcement says people now create 4 billion images per week across ChatGPT, Codex and the API, or more than half a billion per day. The company frames 2.6 around faster iteration and more control over editing rounds, keeping characters, products and brand details consistent.

The Verge notes that before this update, ChatGPT could not natively generate GIFs in a single step.

OpenAI Images 2.6 release date and availability

Images 2.6 launched November 4, 2026. OpenAI says it is available globally across all subscription tiers in ChatGPT and Codex, on desktop, mobile and web. The two new models are offered to developers in the API.

An earlier Facebook group post guessed at an October 29 release. That date was wrong. Per-tier usage caps have not been published in the material available.

Higgsfield said the same day that GPT Image 2.6 is live on its platform, with unlimited use for 7 days for new users.

GPT-Image-2.6 Flare vs Sunburst: API models

OpenAI offers two API models: GPT-Image-2.6 Flare for speed and GPT-Image-2.6 Sunburst for detail and consistency. At low quality, Flare can generate an image in as little as six seconds. Sunburst targets medium and high quality settings.

Two glass lenses, one emitting a light streak and one a sunburst

Per OpenAI, Flare also improves layout quality for posters, product imagery and other professional visuals. Sunburst keeps details more consistent from one edit to the next, according to @OpenAIDevs as quoted in a post from @dotey.

Arena data backs the split. LMArena says Sunburst (Medium) is #2 in Text-to-Image with 1409 points and Flare (Medium) is #4 with 1384, and that Sunburst is #1 in Cartoon, Anime and Fantasy. It also says the models run almost 2x faster than GPT Image 2.5. DesignArena separately reports Sunburst ranked #1 on its Image Arena.

Reception is mixed. One critic on X, @pigeon__s, compared Sunburst output to a model from 349 days earlier and said the new image looks overprocessed. That is a single user's view, not a benchmark.

OpenAI Images 2.6 key features and specs

The main additions are GIFs, reference images, image markup, SVG export and a complexity parameter. Most of them target teams that reuse and revise the same assets many times.

• GIFs: turn a prompt, sketch or image into a loop. Adjust speed and edit individual frames. OpenAI cites animated mascots, reactions, memes and animated storyboards.
• Reference images: save references once and reuse them, per Higgsfield's summary of the feature.
• Image markup: already in ChatGPT, now in the API, so teams can pinpoint where an edit should happen.
• SVG export: designs ready for websites, apps and print.
• Complexity parameter: sets how much planning the model does. Lower settings suit quick edits. Quality controls detail and fidelity separately.

OpenAI has not published GIF resolution, frame rate, duration or aspect ratio limits in the excerpt available. OpenAI's examples include a 6-frame sprite sheet prompt that asks for a looping GIF.

OpenAI Images 2.6 pricing and free access

OpenAI's announcement excerpt does not list API prices. @testingcatalog, which tracks AI releases, posted pricing for image-to-image and text-to-image of $8/$5 input and $30 output, without stating units. Treat it as unverified until OpenAI's pricing page confirms it.

In ChatGPT, Images 2.6 is available on all subscription tiers. Whether free users face tighter limits is not stated.

How to use Images 2.6 for AI video and static ads

Use Flare for quick concept rounds and Sunburst for final static assets where details must survive several edits. That follows OpenAI's own positioning.

A desk with product shots, a tablet showing animation frames and a vector pen

• Static ads: keep a saved product reference, generate variations, then mark up the spot to fix. See our guide to making static ads with AI and the Creative Generator.
• Logos and graphics: SVG export suits elements that need to scale or be recolored.
• Motion: GIFs fit email, landing pages and lightweight social. For paid video hooks, a video model is still the tool. See hook rate and hold rate.
• Comparison: run the same brief across models in the AI Playground.

Test one product, one reference and the same prompt across Flare and Sunburst, then compare edit drift over three rounds.

What to watch next

Watch for confirmed per-image API pricing, GIF output specs and per-tier ChatGPT limits. Until those are public, it is hard to say whether Images 2.6 fits a high-volume ad pipeline.

Also watch independent comparisons on edit consistency. Arena rankings measure preference, not whether a product label survives five edits. Teams chaining image models into repeatable ad pipelines can try that in the Flow Builder once API access is wired into their stack.