GPT Image 2.6 Guide: Features, Price and Prompts

GPT Image 2.6 is OpenAI's newest image model, built for dense text, busy layouts and repeated edits: in Flick it makes an image from a prompt and up to 10 reference images for 5 credits at its default settings.

An ink drawing of a hand holding a drafting compass over a small framed picture of mountains

OpenAI released GPT Image 2.6 on November 4, 2026, with two new controls: Planning, which sets how much the model thinks about a layout before it draws, and output as a transparent cutout or a scalable SVG. For this GPT Image 2.6 guide we ran nine tests in Flick, from a film call sheet and a Korean café menu to a storyboard, a logo and a character kept the same across two edits. Below are its features, prices, the exact prompts and what came back.

How do you use GPT Image 2.6 in Flick?

Add an image node, pick GPT Image 2.6, write the prompt with every word you want shown in quotes, set Planning higher for dense text, and run it for 5 credits at the defaults.
  1. 1Pick GPT Image 2.6

    Add an image node on a Flick canvas, choose GPT Image 2.6 in the model list, and set the aspect ratio of the frame you need.

  2. 2Set Planning

    Leave Planning on medium for most images, raise it to high for tables, menus and pages of text, and drop it to low for small, quick edits.

  3. 3Choose the output

    Turn on Transparent Background for a cutout prop, or SVG (Vector) for a logo or icon you can scale, then run it.

What is GPT Image 2.6?

GPT Image 2.6 is an image generation and editing model from OpenAI, released with a faster sibling, Flare, and a Planning control for dense text and complex layouts.
GPT Image 2.6 in Flick, from the app's model config on November 9, 2026, OpenAI's API docs and our own test files
GPT Image 2.6 in Flick
MakerOpenAI
ReleasedNovember 4, 2026
Ways to use itImage node, Text to Image and Edit Image
Price at default5 credits
Resolutions1K, 2K or 4K
Default 16:9 image, measured1536 × 864
PlanningLow, medium or high, medium by default
OutputPicture, transparent cutout or SVG
Reference imagesUp to 10
Fast versionGPT Image 2.6 Flare
Animated versionGenerate GIF, from a prompt or an image
PlanFree
Where these figures come from

Prices, settings, reference limits and the plan are read from Flick's model config; the image size is measured on our test files. The models and their controls are described in OpenAI's image generation docs.

  • OpenAI ships GPT Image 2.6 as two models: Flare for fast iteration and Sunburst, the one Flick calls GPT Image 2.6, for detail and editing precision.
  • Generate GIF in Flick animates a prompt or a connected image into a short loop of 4, 9 or 16 frames; we have not tested it for this guide.

How much does GPT Image 2.6 cost in Flick?

GPT Image 2.6 costs 5 credits an image at its default Prompt Weight and 1K, rising with weight and size; Planning, transparent backgrounds and SVG add nothing.
GPT Image 2.6 prices in Flick, in credits, from the app's model config on November 9, 2026
Prompt WeightAt 1KAt 2KAt 4K
Medium (default)5914
High101828
Max4070110
  • Free to try: GPT Image 2.6 is on Flick's Free plan, and a new account gets 100 credits.
  • Edits cost the same: an edit in Edit Image is 5 credits at the defaults.
  • A GIF: Generate GIF costs 14 credits for the default 9 frames.

What did GPT Image 2.6 make in our tests?

In our tests GPT Image 2.6 spelled every quoted word and number correctly, in English and Korean, and returned a clean cutout and a real SVG file, though it added small extras the prompts never asked for.

Each card shows the prompt exactly as it was run, at the default Prompt Weight and 1K, with no reference image or style. The call sheet ran with Planning on high, the transparent camera with the cutout switched on and the logo with SVG on.

A printed call sheet for The Last Lighthouse with a four-row table of scenes, locations, cast and call times

A film call sheet, Planning high

Every heading, name and time came back exact, with the four rows ruled into a clean table and the weather box and hospital line where the prompt put them.

A one-page film call sheet for a short film, printed on white paper and laid flat. Header: "THE LAST LIGHTHOUSE" and "CALL SHEET · DAY 3 OF 5". A table with columns "SCENE", "LOCATION", "CAST", "CALL" and four rows: "12 | Cliff path | Mara, Tom | 06:30", "14 | Lamp room | Mara | 09:00", "15 | Keeper's kitchen | Tom, Ada | 11:15", "21 | Harbour wall | All | 15:45". A weather box reads "Sunrise 06:52 · Wind 18 mph". Footer: "Nearest hospital: St. Brannoc's, 4 miles". Clean sans-serif type, thin black rules.
Flick team · GPT Image 2.6 tests
A chalkboard outside a Seoul bakery reading 새벽 베이커리 and Dawn Bakery with four items and prices in won

A Korean café menu

It wrote "새벽 베이커리" and all four prices correctly in chalk, and added a street sign and a window sticker the prompt did not ask for.

A café menu chalkboard on a wooden easel outside a small bakery in Seoul. Hand-lettered in white and yellow chalk: title "새벽 베이커리" with "Dawn Bakery" under it, then four lines with prices: "Croissant  3,500원", "Matcha latte  5,000원", "Strawberry cake  6,500원", "Iced americano  4,000원". Small chalk drawings of a croissant and a coffee cup. Morning light.
Flick team · GPT Image 2.6 tests
A pencil storyboard of six panels: a boy finds a robot in a junkyard and takes it home to bed

A six-panel storyboard

It drew six pencil panels in two rows and typed all six labels letter for letter, dashes and shot sizes kept, with the robot's look held from panel 3 to 6.

A storyboard sheet of six panels in two rows of three, drawn in pencil with grey marker shading on white paper. A boy finds a stray robot in a junkyard: 1 wide shot of the junkyard at dusk, 2 the boy lifting a sheet of metal, 3 close-up of a glowing robot eye, 4 the robot sitting up, 5 the two walking home, 6 the robot asleep on the boy's bed. Under each panel a typed label: "1 WIDE - EXT. JUNKYARD", "2 MED - THE FIND", "3 CU - EYE OPENS", "4 MED - WAKING", "5 WIDE - WALK HOME", "6 MED - GOODNIGHT".
Flick team · GPT Image 2.6 tests
A full-body 3D animated young woman mechanic with goggles, an olive flight jacket with a fox patch and a large wrench

A character for an animated film

Every detail in the prompt is there, from the goggles and fox patch to the grease on one cheek, on a plain background ready to use as a reference.

Full-body character portrait of a young woman mechanic in her twenties for an animated film: short black hair in a messy bob, round goggles pushed up on her head, an oversized olive flight jacket with a patch of a fox on the sleeve, grease on one cheek, holding a large wrench over her shoulder. Clean soft lighting on a plain light grey background, stylized 3D animation look.
Flick team · GPT Image 2.6 tests
A vintage 16mm film camera with a wooden pistol grip and a three-lens turret, cut out with no background

A prop with a transparent background

The camera came back as a cutout with no background, ready to drop into a scene as a prop, though it stamped an invented maker's plate on the body.

A vintage 1960s 16mm film camera with a wooden pistol grip, a three-lens turret and a brass film counter, three-quarter view, studio product lighting, the whole camera in frame.
Flick team · GPT Image 2.6 tests
A vector logo of a navy letter P with an orange folded-paper crescent moon and the name Paper Moon Pictures

A studio logo as SVG

It returned a real SVG file with named layers, but set the studio name from the centre of the canvas so it runs off the right edge, and added a camera shape inside the P.

A flat vector logo for a small film studio called "Paper Moon Pictures": a crescent moon cut from folded paper sitting on the letter P, two colours only, deep navy and warm orange, on white. Clean geometric shapes, the name set in a simple sans-serif under the mark.
Flick team · GPT Image 2.6 tests

What does Planning do in GPT Image 2.6?

Planning sets how much GPT Image 2.6 works out the layout before it draws: high for dense text and busy pages, low for quick, tight edits, at the same price either way.

We ran the call sheet prompt again with Planning on low. It also came back with every word right, and drew bolder type and a sunrise icon on its own. On a page this size both settings held the text, so save high for menus, infographics and pages with more to fit.

The same call sheet made with Planning on low, with bolder type and a small sunrise icon by the weather line
The call sheet prompt with Planning on low: the same words, a looser designFlick team · GPT Image 2.6 tests · GPT Image 2.6
A one-page film call sheet for a short film, printed on white paper and laid flat. Header: "THE LAST LIGHTHOUSE" and "CALL SHEET · DAY 3 OF 5". A table with columns "SCENE", "LOCATION", "CAST", "CALL" and four rows: "12 | Cliff path | Mara, Tom | 06:30", "14 | Lamp room | Mara | 09:00", "15 | Keeper's kitchen | Tom, Ada | 11:15", "21 | Harbour wall | All | 15:45". A weather box reads "Sunrise 06:52 · Wind 18 mph". Footer: "Nearest hospital: St. Brannoc's, 4 miles". Clean sans-serif type, thin black rules.

Does GPT Image 2.6 keep a character the same across edits?

In our test it did: from one portrait, two edits in a row kept her face, hair, goggles, jacket and fox patch while the setting, light and pose changed.

We connected the character portrait as a reference and asked for a new scene, then connected that result and asked for another, each with Planning on low.

The same woman mechanic standing in a rainy aircraft hangar at night beside a small yellow propeller plane

Edit 1: a rainy hangar

Her face, hair and the fox patch carried over unchanged, and the wrench stayed on her shoulder in the new night scene.

The same young woman mechanic from the reference, same face, hair, goggles, olive flight jacket with the fox patch and wrench, now standing full body in a rainy aircraft hangar at night beside a small yellow propeller plane, wet floor reflecting the hangar lights.
Flick team · GPT Image 2.6 tests
The same woman sitting on the wing of a yellow propeller plane at sunrise, eating a sandwich, a wrench beside her

Edit 2: sunrise on the wing

Two edits from the original, she still matches it, and the yellow plane from the first edit came back with the same stripes.

The same woman from the reference, unchanged face, hair, goggles and jacket, now sitting on the wing of the same small yellow propeller plane on a grass airfield at sunrise, eating a sandwich, the wrench beside her.
Flick team · GPT Image 2.6 tests

How do you write a good GPT Image 2.6 prompt?

Quote every word that must appear, say where each goes, rule out the extras you do not want, and match Planning to how much the image has to fit.
  1. 1. Quote every word you need

    Put titles, labels and prices in quotes and say where each goes, as the call sheet did row by row, because each quoted line landed where it was asked.

  2. 2. Rule out extra text

    Write "nothing else written" when a poster or prop should carry your words alone, because GPT Image 2.6 added a street sign to the menu and a maker's plate to the camera.

  3. 3. Raise Planning for dense pages

    Use high for tables, menus and infographics, and low for small edits, since Planning changes the thinking and not the price.

  4. 4. Check SVG text placement

    Open an SVG before you use it and check the type sits inside the canvas, because our logo's name ran off the edge.

  5. 5. Keep a plain reference

    Make a character on a plain background first and connect it to every edit, which is how our mechanic stayed the same across two scenes.

  6. 6. Name what stays the same

    In an edit prompt, list what must not change, such as the face, hair and costume, before you describe the new scene.

How does GPT Image 2.6 compare with GPT Image 2.5?

GPT Image 2.6 adds Planning, transparent backgrounds and SVG output, and costs 5 credits at its default against 5 for GPT Image 2.5.
GPT Image 2.6, its Flare version and GPT Image 2.5 in Flick, from the app's model config on November 9, 2026
GPT Image 2.6GPT Image 2.6 FlareGPT Image 2.5
Credits at default555
PlanningYesYesNo
Transparent backgroundYesYesYes
SVG outputYesYesNo
Reference imagesUp to 10Up to 10Up to 10
PlanFreeFreeFree

For GPT Image 2.5 on its own, see the GPT Image 2.5 guide.

GPT Image 2.6 FAQ

What is GPT Image 2.6?

GPT Image 2.6 is OpenAI's image generation and editing model, released on November 4, 2026, with a Planning control for dense text and layouts. In Flick it makes and edits images at 1K, 2K or 4K from a prompt and up to 10 reference images.

How much does GPT Image 2.6 cost?

In Flick, GPT Image 2.6 costs 5 credits an image at its default Prompt Weight and 1K, and 10 at high weight. Planning, transparent backgrounds and SVG output add nothing.

Is GPT Image 2.6 free?

GPT Image 2.6 runs on Flick's Free plan, and each image uses credits. A new Flick account gets 100 credits, enough for several GPT Image 2.6 images.

What is GPT Image 2.6 Flare?

Flare is the fast version of GPT Image 2.6, made for quick drafts and iterations. In Flick it has the same settings and costs 5 credits at its defaults.

Can GPT Image 2.6 make a transparent background?

Yes. Turn on Transparent Background and it returns the subject as a cutout PNG. Our film camera test came back with no background, ready to use as a prop.

Can GPT Image 2.6 make an SVG?

Yes, from a prompt alone. Turn on SVG (Vector) and it returns a vector file. With a reference image or a Flickpedia style connected, it returns a PNG instead.

Can GPT Image 2.6 make a GIF?

Yes. Generate GIF in Flick uses GPT Image 2.6 to animate a prompt or a connected image into a looping GIF of 4, 9 or 16 frames, for 14 credits at the default 9.

Sources

  • Image prompting guide, OpenAI API docs, checked 2026-11-09.
  • GPT Image, Wikipedia, for the release date, checked 2026-11-09.
  • Prices, settings and limits: Flick's model config, checked 2026-11-09.
  • Test images made in Flick, November 2026.

More on AI images in Flick

Make your first GPT Image 2.6 image in Flick

Pick GPT Image 2.6 on an image node, paste a prompt from this guide, and keep the result beside the rest of your film.

Make a GPT Image 2.6 image in Flick

Try for free, no credit card required.

Ask AI about this page:ChatGPTPerplexityClaude