Skip to content
GPT Image 2.6

New · released Nov 4, 2026

GPT Image 2.6 Generator Online

Create images or edit your photos with Flare or Sunburst, right in your browser. No OpenAI account or API key needed.

Try a prompt:

30 free credits · Standard 1K image = 15 credits · Failed images refunded

An autumn travel poster for Kyoto, with “KYOTO” and “TAKE THE SCENIC ROUTE” in clear lettering…
Illustrated autumn travel poster of Kyoto with traditional houses, maple trees and English headlines
A little idea, a finished image.GPT Image 2.6 Flare · 1024 × 1536

A few ideas to borrow

What does GPT Image 2.6 look like?

Eight examples, from product photos to paper-cut illustration. Open a prompt and make it your own.

Illustrated autumn travel poster of Kyoto with traditional houses, maple trees and English headlines

Travel poster

GPT Image 2.6 Flare · 1024 × 1536

Read the prompt

Use case: ads-marketing. Create a finished vertical travel poster celebrating Kyoto in autumn. A winding narrow lane, traditional wooden machiya, a small stone bridge and vivid maple trees lead the eye toward distant blue-green mountains. Hand-painted gouache and watercolor on warm ivory paper, visible pigment texture, an elegant mid-century travel-poster composition, clay red, ochre and muted teal. Give the landscape generous visual space and use a clear typographic hierarchy. Render these English words exactly, with no other writing: large title at the top "KYOTO", smaller subtitle "TAKE THE SCENIC ROUTE", and bottom caption "AUTUMN IN JAPAN". All lettering must be readable, correctly spelled and integrated into the artwork. Full-bleed finished artwork with a narrow paper margin; no framed mockup, no watermark, no logos.

Dark olive-green olive oil bottle with a brass cap and cream label on a travertine plinth

Product studio

GPT Image 2.6 Flare · 1536 × 1024

Read the prompt

Use case: product-mockup. Make a photorealistic premium ecommerce studio photograph of one dark olive-green glass bottle of extra virgin olive oil, with a brushed brass screw cap and a small cream paper label. The exact label text is "OLIVE" and, beneath it, "EXTRA VIRGIN". Place the bottle on a low travertine plinth against a warm cream seamless backdrop. A single olive twig rests beside the plinth. Soft directional studio light from the upper left reveals realistic glass reflections, a subtle liquid line, stone pores and crisp label printing. Three-quarter front view, the entire bottle in frame, clean balanced composition with breathing room and a soft grounded shadow. Refined commercial photography, sharp product edges, natural materials. No extra bottles, no invented badges, no watermark, no unrelated text.

Original unbranded matte teal water bottle with a dark cap and carry loop on a white studio background
Before
Matte teal water bottle on a travertine table at a sunlit desert cafe terrace
After

A change of scene

GPT Image 2.6 Sunburst · 1536 × 1024 · edit

Read the prompt

Use case: precise-object-edit. Input image 1 is the edit target. Replace the white studio background and white supporting surface with a sunlit outdoor cafe terrace in a desert landscape. Put the exact same teal water bottle on a warm pale travertine table; behind it, show softly defocused terracotta architecture, desert hills and a small patch of muted sage foliage. Warm late-afternoon light from the upper left, believable soft contact shadow and subtle environmental reflections. Preserve the product's identity: identical matte teal color, cylindrical body, shoulder and neck geometry, ribbed charcoal screw cap, small arched carry loop, size in the frame, front three-quarter camera angle and position. Keep the bottle fully in frame and sharply focused. Change only its surroundings and the lighting needed to integrate it naturally. No added label, logo, text, extra bottle, hands or watermark. Deliver one finished photo, not a before-and-after collage.

Three-step infographic titled FROM IDEA TO IMAGE with writing, reference and download icons

An idea, explained

GPT Image 2.6 Sunburst · 1536 × 1024

Read the prompt

Use case: infographic-diagram. Design a polished landscape editorial infographic explaining a simple three-step creative workflow. Warm ivory paper background, ink-dark type, restrained terracotta, dusty blue and sage accents, precise grid, broad margins, generous whitespace. Title at top, exactly: "FROM IDEA TO IMAGE". Below, arrange exactly three equally sized panels from left to right, connected by simple arrows. Panel 1 has a pencil-and-note icon, number "01", heading "WRITE A PROMPT", caption "Describe the scene". Panel 2 has a framed-photo icon, number "02", heading "ADD A REFERENCE", caption "Guide the details". Panel 3 has a downward-download icon, number "03", heading "GENERATE & SAVE", caption "Review your image". Render every quoted phrase exactly and legibly with coherent typography. Use small handcrafted paper-cut illustrations to support the process. No charts, no invented statistics, no extra steps, no logos, no watermark. Finished flat infographic, not a physical mockup.

Fictional woman ceramic artist with short dark hair and an ochre apron in a pottery studio

Natural portrait

GPT Image 2.6 Sunburst · 1024 × 1536

Read the prompt

Use case: photorealistic-natural. Create a candid editorial portrait of a fictional adult woman ceramic artist in her thirties in a working pottery studio. Chest-up composition at eye level, her body turned slightly toward the workbench and her face turned naturally toward the camera, a quiet relaxed expression. She has short wavy dark hair, warm brown eyes, natural skin with visible pores and a few freckles, a cream linen shirt and an ochre work apron lightly dusted with clay. Soft overcast window light from the left, gentle shadow falloff, shelves of handmade bowls and clay tools blurred in the background. Realistic 85mm portrait-lens rendering, shallow depth of field, warm restrained colors, tactile linen and clay, thoughtful contemporary magazine photography. A fictional person, no celebrity resemblance. No lettering, no logos, no watermark, no beauty-filter skin.

Six gouache stickers: coffee cup, cat, camera, sun, mushroom and potted olive plant

Sticker collection

GPT Image 2.6 Flare · 1536 × 1536

Read the prompt

Use case: illustration-story. Create a charming finished sticker sheet with exactly six separate die-cut illustrations, arranged in a tidy two-by-three grid on a solid warm cream paper background. The six subjects are: a smiling terracotta coffee cup, a sleepy sage-green cat, a tiny dusty-blue camera, an ochre sun with rounded rays, a cheerful red mushroom, and a small potted olive plant. Each sticker has a distinct clear silhouette, a broad ivory die-cut border and a tiny soft paper shadow; leave ample empty space between stickers so none overlap or touch. Friendly hand-painted gouache illustration with visible brush grain, rounded shapes, ink-brown fine details and a cohesive muted palette. No words, numbers, labels, logos or watermark. Opaque background throughout. Flat front-facing artwork, no hands, no packaging mockup.

Layered paper-cut terracotta cottage beside a blue river with sage leaves and ochre flowers

Paper-cut world

GPT Image 2.6 Flare · 1536 × 1536

Read the prompt

Use case: stylized-concept. Create a richly tactile paper-cut illustration of a small terracotta cottage with a cream roof at the edge of a winding dusty-blue river, surrounded by layered sage leaves, ochre meadow flowers and rounded distant hills. A small ochre sun sits above the hills. Build the scene from overlapping sheets of textured colored paper with clean hand-cut contours, visible paper fibers, slight variations at the edges and soft directional shadows that reveal the shallow physical layers. Balanced square composition with the cottage as the focal point and a playful river curve leading toward it. Warm ivory background, clay red, dusty sky blue, sage green, ochre and deep ink brown; gentle handmade editorial mood. View nearly straight down onto the layered artwork, no photoreal landscape. No text, no logos, no watermark, no frame or staged desk props.

Video thumbnail with a fictional creator, three art prints and the headlines ONE PROMPT and BIG IDEAS

A thumbnail that reads

GPT Image 2.6 Sunburst · 1536 × 864

Read the prompt

Use case: ads-marketing. Create a finished landscape YouTube thumbnail about transforming an image idea into visual artwork. At the left, a fictional adult male creator with short dark hair and a rust-colored shirt looks toward the right with a naturally impressed expression, one open hand gesturing toward three floating art prints. At the right, the prints show a colorful mountain travel illustration, a clean ceramic mug product photo and a bold botanical paper-cut artwork. Warm cream background with dusty-blue and terracotta graphic accents, clear foreground separation, polished editorial photography mixed with crisp graphic layout. Across the upper center, place very large bold dark lettering with a light outline, exactly "ONE PROMPT". Beneath it, a smaller terracotta label reads exactly "BIG IDEAS". Keep both lines readable at small thumbnail size, away from the face and prints. No other text, no logos, no interface screenshot, no fabricated metrics, no watermark.

Meet the new edition

What’s new in GPT Image 2.6?

The November 4 release adds a choice of two models. Here is what you can use in this generator.

Two ways to work

Flare favors speed. Sunburst favors precision and consistency across edits. Choose the version for the image you are making.

A faster starting point

OpenAI reports 50% faster image generation for the new release. This is OpenAI’s claim, rather than a timing guarantee for this service.

Plan the composition

Choose low, medium or high composition planning in the generator. Planning is a separate control from output quality and does not change the credit cost.

Start with your images

Use reference images to guide a new scene or edit an existing photo. Describe what should change and what needs to stay the same.

Read OpenAI’s announcement

Pick your working style

Flare or Sunburst — which should you use?

The choice is about your task. Matching settings cost the same with either model.

For a quicker first draft

GPT Image 2.6 Flare

Start with Flare when you want to try an idea, compare directions or make several versions before choosing one.

A good starting point for social graphics, product concepts and everyday image generation.

For detail and repeat edits

GPT Image 2.6 Sunburst

Try Sunburst when the brief needs careful detail or when you want consistency through a series of edits. OpenAI highlights its consistency at medium and high quality.

Useful for refined product scenes, detailed compositions and editing a reference image.

Same credits for both — switch per image.

The practical details

Why make your images here?

Both models, reference editing and a price you can check before you start.

No Sunburst surcharge

Flare and Sunburst use the same credit table. Change your model without paying extra just for the name.

Up to 16 reference images

Add your own photos or product references. Uploading an image switches the generator into editing mode automatically.

Choose the finish, see the cost

Five quality levels, 1K, 2K or 4K output, and composition planning. Your credit total is shown before you generate.

Room for a first try

Sign up for 30 credits, enough for 2 standard 1K images. Failed generations are refunded automatically. No OpenAI account or API key needed.

From a sentence to an image

How to use GPT Image 2.6

  1. Describe the image

    Name the subject, scene, light and style. Put any words you want in the picture in quotation marks, and say where they belong.

  2. Choose your settings

    Pick Flare or Sunburst, an aspect ratio, resolution and quality. Add reference images if you want to edit a photo or guide the result.

  3. Check, generate, download

    Review the credit total, then generate. Check the image and any lettering before downloading or making your next edit.

Start with your idea

See the credit cost first. Failed generations are refunded automatically.

One model, plenty of possibilities

What can you make with GPT Image 2.6?

Start with a concrete brief: a product, a message, a scene or an idea to explain.

Product photography

Try a clean studio shot or a styled product scene. Use a reference to guide the shape, materials and identifying details.

Posters and ads

Combine an image with a headline. Describe the hierarchy and quote the exact words, then check the lettering in your result.

Social posts and thumbnails

Make a focused visual for a feed, cover or video thumbnail. Keep the subject clear and plan for the crop you need.

Photo and scene edits

Change the setting, lighting or styling around your reference. Tell the model which details should remain unchanged.

Infographics and UI sketches

Turn a process into a visual sequence, or explore an interface idea. Review all labels, numbers and factual content yourself.

Editorial art and stickers

Explore a paper-cut scene, a small illustration or a sticker collection. Specify a shared palette and style to keep the set coherent.

A small table, no guesswork

How much does an image cost?

A standard medium-quality 1K image costs 15 credits. Flare and Sunburst cost the same at matching settings.

Editing includes one reference in the table below. Each additional reference adds 3 credits. Composition planning does not change the price.

See full pricing
Medium quality · credits per image
ResolutionText to imageEdit · 1 reference
1K15 credits18 credits
2K26 credits29 credits
4K45 credits48 credits

Both models use this table. The generator shows your total before you start.

A few useful answers

GPT Image 2.6 FAQ

Free credits, editing, model choices and what is available here.

What is GPT Image 2.6?

GPT Image 2.6 is OpenAI’s image model release from November 4, 2026. Its API has two variants, Flare and Sunburst. This service lets you use both in your browser to create images from text or edit images with references. The model ID is gpt-image-2.6.

Can I use GPT Image 2.6 for free?

Sign up to receive 30 free credits, enough for 2 medium-quality 1K text-to-image generations at 15 credits each. No card is required to try them. Higher quality, larger images and reference editing use more credits.

Flare or Sunburst — which should I choose?

Start with Flare for quick drafts and exploring ideas. Try Sunburst for precise details and consistency across repeated edits; OpenAI highlights this at medium and high quality. Both cost the same here when quality, resolution and reference count match.

How much does one GPT Image 2.6 image cost here?

A medium-quality 1K image from text costs 15 credits. The total depends on quality, resolution and reference count. Each extra reference after the first adds 3 credits. Flare and Sunburst share the same prices, and composition planning does not add a charge. Check the total in the generator before starting.

See full pricing
Can I edit my own photos with GPT Image 2.6?

Yes. Upload up to 16 reference images to switch automatically to image-to-image editing. Describe the changes you want and the details that should stay the same, such as the subject, product shape or composition. Review the result before using it.

Does GPT Image 2.6 render readable text in images?

Yes, you can ask for headlines, labels and other lettering. Put the exact text in quotation marks and specify its position and relative size. Always check spelling and layout in the finished image; text accuracy is not guaranteed.

Can I create GIFs or SVG files here?

Not here yet. OpenAI added GIF support to Images 2.6, and third-party reports also mention SVG. This service currently outputs still PNG, JPEG or WebP images. GIF, SVG, mask or markup editing, and transparent-background controls are not available here; new capabilities will be added when supported by this integration.

Is GPT Image 2.6 available in ChatGPT? Do I need an OpenAI account here?

In ChatGPT the release is called ChatGPT Images 2.6. You do not need an OpenAI account or your own OpenAI API key to use this service. Sign in to this site and use its credits to generate or edit images.

Can I use the images commercially?

You can use generated images for commercial projects subject to the service terms and applicable OpenAI policies. You are responsible for the rights to your uploads and for checking the output, including trademarks, likenesses and other third-party rights. AI-generated images may not be unique.

Read the terms
Is this the official OpenAI website?

No. This is an independent third-party service, not the official OpenAI website.

Is there a GPT Image 2.6 API?

Yes. In your account, go to Developers → API Keys to create a key for this service. Use the model ID gpt-image-2.6 and consult the API documentation for request parameters and examples.

Your next idea starts here

What will you make first?

Bring a sentence or a reference photo. Choose your model, check the cost and see where the idea takes you.

Try GPT Image 2.6
  • 30 signup credits · no card needed to try
  • Flare and Sunburst at the same price
  • Failed generations refunded automatically