Models
Output format
Images (Optional)
Not sure what to write? Try an example:
Make looping GIFs and editable SVG vectors from one prompt, or turn any photo into menus, flyers and infographics with clean text, up to 50% faster with 4K export.
GPT Image 2.6 is OpenAI's image generation and editing model released on November 4, 2026, the same update that brought ChatGPT Images 2.6. It returns images up to 50% faster than the previous version, plans the layout before it draws, and can now export animated GIFs and SVG vector files. On this page you can start from your own photo, or switch the output format to GIF or SVG and create a looping animation or an editable vector graphic from a prompt alone.
Version 2.6 adds two new output formats and improves the designs that used to fail most often: pages full of text, where names came back misspelled and lines ended up crowded.
Less waiting on each run means you can try a prompt, adjust one line and run it again without losing your train of thought.
Describe a subject and a simple motion, choose 2 to 16 frames and how long each one shows, and get a smooth looping GIF for stickers, banners and loading animations, ready to post or drop into a chat.
Logos, icons and flat illustrations come back as clean vector files you can open in Figma or Illustrator, recolor and scale to any size without blur.
The model works out where the title, the picture and each block of text go before it renders, and words you put in quotes stay spelled the way you typed them.
Pick GIF or SVG above the prompt box and describe what you want. No upload or design software needed.
This shop window animation came from one sentence: a cafe storefront with confetti falling, flat illustration, seamless loop. Set any frame count from 2 to 16 and any frame duration from 20 to 2000 milliseconds; 16 frames at 100 milliseconds each make a smooth loop for social posts, emails and chat stickers.
Ask for a set of icons with the same stroke width and two colors, and you get one SVG file with separate shapes for every icon. Change a color or resize it in any vector editor and it stays sharp.
Three tasks where the faster speed and the layout planning make the most difference.
Upload a photo of a dish or drink and list your items with prices in the prompt. You get a finished board with the photo on top and every line in order, ready for a counter screen or a printout.
Give the model a photo and four or five short points, and it arranges them into a card with a title, icons and labels around the picture, sized for a post or a slide.
Skip the trial-and-error wait: generate fast, low-latency drafts to verify wording, line breaks, and visual hierarchy quickly, then render the approved design in crisp 4K for print or display.
Upload a photo of an astronaut toy and a moon model. This everyday tabletop shot becomes the starting point for a cinematic exhibition poster.
Ask for a cinematic lunar exhibition while keeping the toy’s suit and pose. Specify “BEYOND EARTH” at the top, “A LUNAR EXPERIENCE” below, and “OCT 24–26 · CITY MUSEUM” for the event details.
Check the preserved suit and pose, the lunar scene, and every line of the poster. Refine any wording that needs attention, then download your chosen version.
People who need clear text on an image and do not want to open a design app for it.
Daily specials boards, seasonal menus and price cards made from photos of your own dishes.
Explainer cards, step-by-step visuals and slide images where each label has to be spelled right.
Sale flyers, price tags and product cards with the offer, the price and the dates in the right places.
Turn dish and product photos into menus and flyers, or a single prompt into a looping GIF or an editable SVG, without opening a design app.
Answers about GIF and SVG output, layout planning, text rendering, prompt formulas and commercial use.
