Use GPT Image 2.6 online, then animate it

GPT Image 2.6 is OpenAI's image model, released November 4, 2026 in two versions: Flare, the fast one, and Sunburst, for higher quality and consistent edits. Make product shots, ads and video start frames here, from text or up to 16 reference photos.

  • Pay per image, no subscription
  • Up to 4K
  • Animate any image into a video

GPT Image 2.6New

OpenAI image model

Version
Prompt0 / 20,000
Reference imagesOptional · up to 16 · +1 credit each
Format
Aspect ratio
Resolution
Background

Pay per image, no subscription

Made with GPT Image 2.6

Click one to load its prompt, then generate your own version.

How do I turn a GPT Image 2.6 image into a video?

Open the image and press Animate this image. The video generator opens with it as the start frame, so there is nothing to download or upload again, and the image and the video use the same credits.

  1. Make the start frame

    Generate the scene, product shot or character with GPT Image 2.6. Its exact text and layout make a clean first frame.

  2. Press Animate this image

    The video generator opens in Image to Video with your image as the start frame and its shape matched.

  3. Pick a video model and describe the motion

    Choose Seedance, Kling, MiniMax, Wan or Grok Imagine, write the camera move or action, and generate.

Not sure which video model to pick? See the same start photo on four models in image to video.

What is GPT Image 2.6 good at?

Text inside the image, product shots and edits that keep the details. These were made on VisionMorphic with the prompt under each one; none were retouched.

Folk festival poster titled Pine Hollow Sessions with mountains, a cabin and a five-band lineup
Flare, 2K, 2:3. Every word on the poster came out as written, including the five band names.
Prompt
A screen-printed concert poster for an indie folk festival called 'Pine Hollow Sessions'. Three layered mountain ridges in forest green, mustard and cream, a small cabin with a glowing window, hand-drawn pine trees. Bold condensed headline 'PINE HOLLOW SESSIONS', a line below reading 'August 14-16 · Lakeside Meadow', and a lineup block with five made-up band names in a clean grotesk typeface. Subtle paper grain and slight ink misregistration.
Chalkboard cafe menu titled Morning Light Cafe with coffee and bakery prices
Flare, 2K, 3:4. Eight items and eight prices, all correct.
Prompt
A chalkboard cafe menu photographed straight on, hanging on a white brick wall. Hand-lettered title 'Morning Light Cafe'. Two columns: COFFEE with Espresso 3.00, Flat White 4.25, Oat Latte 4.75, Cold Brew 4.50; BAKERY with Butter Croissant 3.50, Cardamom Bun 4.00, Banana Bread 3.75, Almond Tart 4.25. Small chalk drawings of a coffee cup and a croissant. Every word spelled correctly, neat legible lettering.
Lighthouse keeper in a yellow raincoat walking along a stormy stone pier at blue hour
Sunburst, 2K, 16:9.
Prompt
Cinematic wide shot of a lone lighthouse keeper in a yellow raincoat walking along a wet stone pier at blue hour, waves crashing against the rocks, the lighthouse beam cutting through light rain, warm light from a small cottage window, shot on a 35mm lens, film grain, moody teal and amber color grade. Full-frame image with no black bars and no letterboxing.

Can GPT Image 2.6 edit a photo and keep the label?

Yes. We made a product shot with Flare, then gave it to Sunburst as a reference and asked for a new scene. The bottle, cap and every word on the label stayed the same.

Amber serum bottle labeled LUMA Vitamin C Serum on a travertine block
Step 1 · Flare, 2K, text to image.
Prompt
Studio product photo of a frosted amber glass serum bottle with a matte black dropper cap, standing on a rough travertine block. The label reads 'LUMA · Vitamin C Serum · 30 ml' in a thin modern sans-serif. Soft daylight from the left, a sharp shadow of a palm leaf across the background wall, warm sand-colored backdrop, shallow depth of field, high-end skincare advertising.
The same LUMA serum bottle placed on wet volcanic rocks by the sea at sunset
Step 2 · Sunburst, 2K, with the step 1 image as the reference.
Prompt
Put this exact serum bottle on wet black volcanic rocks at the edge of the sea at golden hour, with a small wave splashing behind it and droplets on the glass. Keep the bottle shape, the frosted amber glass, the black dropper cap and the label text exactly the same.

Want the bottle to move? Use the image as the start frame in image to video with Seedance 2.5 or Kling 4.0.

GPT Image 2.6 Flare vs Sunburst: which should I use?

Start with Flare for posters, menus and product shots; switch to Sunburst for more detail or when you edit the same image several times. Both cost the same. Here is one prompt on each, both at 2K.

Ceramics studio with a sleeping orange cat, made with GPT Image 2.6 Flare
Flare · 19 seconds
Ceramics studio with a sleeping orange cat, made with GPT Image 2.6 Sunburst
Sunburst · 52 seconds
Prompt

A cozy ceramics studio on a sunny afternoon: a potter's wheel with a half-thrown vase, shelves of glazed bowls in speckled blues and creams, dust floating in the light from a large window, a sleeping orange cat on a stool, a small handwritten sign on the shelf reading 'Glaze Day: Thursday'.

FeatureFlareSunburst
Made forSpeed and layout: posters, product shots, menusDetail and edits that stay consistent
Time in our test19 seconds52 seconds
Credits per imageSameSame
Text and referencesPrompt up to 20,000 characters, up to 16 reference imagesSame
Sizes1K, 2K, 4K; 15 aspect ratios plus AutoSame

“Made for” follows OpenAI's announcement. Times are from one generation each on November 6, 2026 and will vary.

What size are GPT Image 2.6 images, and how big can I print them?

A 4K image is up to 3840 pixels on the long side, enough for a 12.8-inch print at 300 ppi or a 25.6-inch poster at 150 ppi. The exact size depends on the aspect ratio; these are the sizes our own generations came back at.

SettingPixelsPrint at 300 ppiPrint at 150 ppi
1K · 1:11024 × 10243.4 × 3.4 in6.8 × 6.8 in
1K · 2:31024 × 15363.4 × 5.1 in6.8 × 10.2 in
2K · 1:12048 × 20486.8 × 6.8 in13.7 × 13.7 in
2K · 2:31360 × 20484.5 × 6.8 in9.1 × 13.7 in
2K · 4:52048 × 25606.8 × 8.5 in13.7 × 17.1 in
2K · 16:92048 × 11526.8 × 3.8 in13.7 × 7.7 in
4K · 2:32336 × 35207.8 × 11.7 in15.6 × 23.5 in
4K · 16:93840 × 216012.8 × 7.2 in25.6 × 14.4 in

How much does GPT Image 2.6 cost?

4 credits for a 1K image, 7 for 2K and 10 for 4K, plus 1 credit per reference image. Flare and Sunburst cost the same, and a one-time pack of 250 credits for $4.99 makes about 62 images at 1K with no subscription.

One 1K image
4 credits
One 2K image
7 credits
One 4K image (Basic and higher)
10 credits
Each reference image
+1 credit
Animated GIF, 512 × 512
4 credits up to 8 frames, 7 for 9–16
Transparent background, JPEG or WEBP
No extra cost (transparent is 1K only)
Flare or Sunburst
Same price
GIFs
Paid plan or purchased credits
One-time credit pack
250 credits for $4.99, about 62 images at 1K
Visible watermark
None, on every plan
Storage on Free
7 days
Commercial use
Basic, Pro and Ultra; not on Free
Free trial
1 image a day at 1K or 2K with the free credits; resets at 00:00 UTC

Credits as of November 6, 2026. Purchased credits are valid for 12 months and work with every image and video model.

How do I use GPT Image 2.6 online?

  1. Sign in and add credits

    Sign in with Google, no card needed. Credit packs start at 250 credits for $4.99, with no subscription.

  2. Write a prompt or add photos

    Describe the image, including any words it should show. To edit, restyle or combine, add up to 16 reference images.

  3. Generate, then download or animate

    Pick Flare or Sunburst, the shape and 1K, 2K or 4K. The button shows the cost first, failed images return their credits, and any result can go straight to video.

Questions

Can I turn a GPT Image 2.6 image into a video?

Yes, in one click. Open the image and press Animate this image: the video generator opens with it as the start frame, on the same account and credits. Pick Seedance, Kling, MiniMax, Wan or Grok Imagine and describe the motion. See how image to video works.

Which video models can animate a GPT Image 2.6 image?

Seedance 2.5 and 2.0, Kling 4.0 and 3.0, MiniMax H3, Wan 3.0 and Grok Imagine 1.5 all take it as the start frame. Kling 4.0 and MiniMax H3 always add sound, Seedance 2.5 goes up to 30 seconds, and the generator shows each video's credit cost before you submit. Compare the video models on the same prompts.

Can I use GPT Image 2.6 without ChatGPT Plus?

Yes. On VisionMorphic you pay per image with credits: 4 credits at 1K, from a one-time pack of 250 credits for $4.99, valid for 12 months. There is no daily cap on paid credits, and no ChatGPT account or subscription is needed.

Can I use GPT Image 2.6 images commercially?

Yes on Basic, Pro and Ultra, for ads, product listings and items you sell. Images made on the Free plan are for personal use only.

Can GPT Image 2.6 edit my own photos?

Yes. Add up to 16 JPG, PNG or WEBP images as references and describe the change, for example a new background for a product with the label kept the same. Each reference image adds 1 credit.

What is the difference between GPT Image 2.6 Flare and Sunburst?

OpenAI made Flare the fast version, aimed at layouts such as posters and product images, and Sunburst the higher-quality version with more consistent edits. On VisionMorphic both take the same inputs and cost the same credits. In our test with the same prompt, Flare finished in 19 seconds and Sunburst in 52.

How big can I print a GPT Image 2.6 image?

A 4K image in 16:9 is 3840 × 2160 pixels: 12.8 × 7.2 inches at 300 ppi for close viewing, or 25.6 × 14.4 inches at 150 ppi for a poster seen from a few feet away. 2K and 1K images print proportionally smaller.

Do GPT Image 2.6 images have a watermark?

There is no visible watermark on any plan, including Free. Like every OpenAI image, the PNG file carries C2PA Content Credentials, invisible metadata that says it was made with OpenAI; we checked this on our own outputs.

Can I use GPT Image 2.6 for free?

Yes, as a daily free trial: you can try 1 GPT Image 2.6 image free every day at 1K or 2K with the free credits, and it resets at 00:00 UTC. With a credit pack (from 250 credits for $4.99) or a paid plan there is no daily limit and GIFs open up; 4K needs a paid plan.

Is this the official OpenAI website?

No. VisionMorphic is an independent product that runs GPT Image 2.6 through an API provider. It is not affiliated with or endorsed by OpenAI.

Make the image, then make it move

Pay per image with no subscription, and send any result to video in one click. Flare and Sunburst are in the same generator.

Try GPT Image 2.6