Combine visual references
Give each image a role: subject, environment, or palette. Resolve conflicting instructions in the prompt so the intended priority is clear.
GPT Image 2.5 · Default generation: 3 points
GPT Image 2.5 Image Generator & Editor
Turn a brief into a campaign visual, explore a poster layout, or rework an image with references. Create with GPT Image 2.5 in one online workspace, with a default cost of 3 points per generation.
01
Try GPT Image 2.5 online
GPT Image 2.5 is selected. Start with text or switch to reference creation to upload images. The generate button shows the current estimate; the default is 3 points.
AI Image
Results
Creation historyGenerated content will appear here
02 · Overview
GPT Image 2.5 at a glance
Last reviewed:
- Model family
- OpenAI GPT Image 2.5
- Related product
- ChatGPT Images 2.5
- Platform model
- gpt-image-2.5
- Inputs
- Text and reference images
- Workflows
- Text to image · Reference creation
- Default cost
- 3 platform points / generation
What is GPT Image 2.5?
GPT Image 2.5 is the image-generation family behind ChatGPT Images 2.5, announced by OpenAI on September 8, 2026. OpenAI highlights improved reference fidelity, more dependable revisions, and more natural lighting and texture. Those changes make image creation and editing useful topics to evaluate together.
The official API has two named models: GPT-Image-2.5 Flare focuses on fast everyday generation, while GPT-Image-2.5 Sunburst prioritizes editing precision. Here, you select the platform entry GPT Image 2.5. This workspace does not currently expose a separate Flare or Sunburst selector.
Begin with an actual deliverable: a product card, launch poster, social post, character study, or presentation graphic. The demo briefs below show what to describe and what to inspect. The gallery includes supplied editing comparisons; the other sections show supplied creative examples.
03 · Capabilities
From a written brief to an image you can refine
Four core workflows, followed by practical controls for your next creative iteration.
Iterate one change at a time
For another revision, upload the approved result as a reference and ask for one specific change. Compare it with the previous image before proceeding.
Know the point estimate
The default GPT Image 2.5 generation costs 3 platform points. Review the button before each submission, including when switching models or workflows.
Reuse a structured prompt
Load one of the examples below, then replace the subject, text, palette, and constraints. Keep the successful brief alongside your downloaded image.
03A · Model comparison
GPT Image 2.5 vs GPT Image 2
A comparison of the two entries available on this platform. Compare the same brief and references; the demo comparisons are not a benchmark.
| What to compare | GPT Image 2.5 | GPT Image 2 |
|---|---|---|
| Platform model | gpt-image-2.5 | gpt-image-2 |
| Starting workflow | Text prompts and reference images | Text prompts and reference images |
| Creative evaluation | Try a new reference edit and inspect the preserved details | Use an existing result as your comparison baseline |
| Visible output controls | Current workspace options; no manual resolution or quality selector | 1K / 2K / 4K and available quality options |
| Point estimate | 3 points by default; check the generate button | Varies with the selected settings and references |
| Best way to decide | Compare text, subject details, composition, and actual turnaround | Keep the version that meets your own delivery requirements |
GPT Image 2.5 pricing on Nano Banana Pro
Start with a clear brief and review the estimate before submitting. This workspace uses Nano Banana Pro points.
View plans and point packs 3 points
Default cost per generation. Additional attempts are separate tasks; the generate button shows the current estimate.
One account, one point balance
Sign in with a Nano Banana Pro account and sufficient points. A ChatGPT subscription does not provide points here.
04 · Use cases
GPT Image 2.5 demo ideas and use cases
Five supplied visual comparisons covering group portraits, outfit changes, pet costumes, photo restoration, and room lighting.
Visual examples01
Multiple references · A shared group photo
Three individual portraits become a group snapshot in a shared setting. Compare facial details, relative proportions, and the lighting across the three subjects.
Visual examples02
Portrait editing · Change an outfit
A red polo shirt is replaced with an ivory tuxedo in a childhood portrait. Inspect the face, clothing edges, and how the new outfit fits the original pose.
Visual examples03
Pet portraits · A creative costume
A dog on a patio gains a racing outfit and helmet. Compare the fur, pose, background furniture, and shadows around the added accessories.
Visual examples04
Photo restoration · From print to portrait
A photograph held in a hand becomes a clean portrait. Compare the removal of the paper border and glare, the framing, and the retained facial details.
Visual examples05
Interior editing · Adjust the lighting
A bedside lamp changes the mood of a bedroom. Compare the warm illumination on the wall and bedding, along with the furniture and room layout.
05 · Prompts
GPT Image 2.5 prompt examples
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Prompt 01
National park stamp collection
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a landscape sheet of eight vintage national park souvenir stamps in two rows of four on a black background. Give every stamp a cream perforated edge, serif park name at the top and a small NATIONAL PARK subtitle. In reading order: Yellowstone with a geyser, Grand Canyon with layered canyon cliffs, Acadia with a coastal lighthouse, Zion with red cliffs, Glacier with an alpine lake, Great Smoky Mountains with misty forest ridges, Denali with a snowy peak and caribou, Everglades with a wetland sunset and heron. Use textured screen-print illustration, muted golden yellow, coral, turquoise and forest green. Keep margins and stamp sizes consistent; render each park name once and do not overlap text with the landscape.
Prompt 02
Classical columns · Editorial scene
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a vertical editorial illustration inspired by classical antiquity. Two adult figures stand on pale stone steps between monumental fluted columns: an armored guard wearing an engraved bronze helmet and long olive-green cloak with a geometric border, beside a woman in a flowing black evening dress. Show both figures full length. A narrow strip of deep blue sky separates the columns; a red runner crosses the bottom steps. Use strong midday shadows, restrained sand, olive, black and navy colors, fine engraved linework and a weathered print texture. Emphasize architectural scale and fabric detail. No text.
Prompt 03
Black cat sticker poster
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a vertical vintage sticker advertisement on a worn cobalt-blue paper background. A small black cat with large golden eyes sits in the lower half and holds a cream sticker sheet containing a goldfish, speech bubbles, stars, a heart and a thumbs-up. Put the oversized bold yellow headline "ChatGPT Stickers" on two lines across the top, followed by the smaller white subtitle "New sticker pack". Add the small Japanese heading "チャットGPT ステッカー" above the headline. Use realistic soft black fur, crisp graphic typography, a restrained blue-yellow palette, faded folds and scuffed cream edges. Keep the whole cat and sticker sheet inside the frame.
Prompt 04
Flare and Sunburst · Mountain sunrise comparison
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a clean side-by-side image comparison divided vertically into two equal panels. Add a small dark translucent label “Flare” in the upper-left corner of the left panel and “Sunburst” in the upper-left corner of the right panel. Both panels show the same alpine valley at sunrise: a clear rocky stream flows from the foreground toward the low sun between two mountains, with green meadow grass, scattered stones, one mature leafy tree near the stream, warm golden reflections in the water, and a blue sky with small glowing clouds. Make the left interpretation bright, lush and softly radiant; make the right interpretation slightly cooler and more dramatic, with snow on the peaks, stronger water texture and a richer orange horizon. Photorealistic landscape photography, detailed natural light, 4:3 composition, no text beyond the two panel labels.
Prompt 05
Flare and Sunburst · Streetwear campaign comparison
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a vertical split comparison with two equal streetwear campaign panels and small dark labels “Flare” and “Sunburst” at the top. In both panels, show a diverse group of five young adult friends in relaxed oversized streetwear sitting together on concrete steps at a skate spot, with skateboards in the foreground and a modern city skyline behind them. Use candid smiles, natural group interaction, golden-hour light and editorial fashion photography. Incorporate the exact brand word “Thread.” in bold handwritten script and the exact tagline “Yours to Create.” in clean sans-serif type. In the left panel, place the white logo near the sky and the tagline across the lower middle; in the right panel, use a black logo and smaller tagline in the upper sky. Keep faces, hands and clothing realistic, preserve readable typography, 4:3 overall composition.
Prompt 06
Flare and Sunburst · Fisherman portrait comparison
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a side-by-side documentary photography comparison divided vertically, with small dark translucent labels “Flare” and “Sunburst” at the upper-left of each panel. Both panels portray an elderly bearded fisherman working on a fishing net aboard a weathered boat, with a calm sea and rocky coast behind him and a friendly working dog nearby. Show authentic wrinkles, rough hands, muted knit clothing, waterproof bib overalls, ropes and nets. In the left panel he wears a dark beanie and green-black overalls while seated closer to camera; in the right panel he wears a dark sailor cap, gray sweater and ochre overalls while standing. Soft overcast coastal light, restrained blue-gray palette, tactile documentary realism, 4:3 overall composition, no text beyond panel labels.
Prompt 07
Flare and Sunburst · Coffee machine infographic comparison
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Design a dense side-by-side technical infographic comparing two visual treatments of the same automatic bean-to-cup coffee machine. Add compact labels “Flare” and “Sunburst” at the top of the corresponding panels. Each panel should contain a realistic cutaway of a premium black and silver coffee machine, numbered internal components, colored flow paths for beans, water, steam and milk, a step-by-step brewing sequence, a component legend and a simplified process diagram. Use the headline “Inside an Automatic Coffee Machine” and the subtitle “From beans to a perfect cup.” Make the left panel bright and encyclopedic on white, with multiple small callouts; make the right panel more polished and product-led, with a dark lower information band. Keep numbering consistent, diagrams aligned and all English labels legible. Landscape 4:3 layout, engineering editorial style.
Prompt 08
Field research equipment · Documentary POV
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Photorealistic documentary point-of-view photograph of a scientific field team unloading rugged black equipment cases from several white SUVs outside a modern earth sciences research building in a mountain valley. The viewer’s gloved hands grip the large hard case in the immediate foreground, creating a strong first-person perspective. Three researchers in practical jackets organize instrument cases, water containers, tripods and a survey antenna across the parking area. Include glass entrance doors, pale stone walls, trees and distant snow-streaked mountains under a bright overcast sky. Wide-angle action-camera perspective, natural colors, crisp environmental detail, believable working moment, 4:3 composition, no watermark or added text.
Prompt 09
Friends’ poker night · Warm lifestyle scene
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a warm candid lifestyle photograph of seven adult friends gathered around a round wooden dining table for a casual poker night in a cozy apartment. Two seated players hold cards in the foreground; chips, playing cards, drinks, a phone and a snack bowl are scattered naturally across the table. At center, a laughing man points across the table while another friend photographs the moment with a phone; a woman carries a bowl of snacks and two others watch the game. Use warm practical lamp light, an intimate lived-in interior, natural expressions, realistic hands and cards, shallow but inclusive depth of field, editorial documentary photography, landscape 4:3 composition, no watermark.
Prompt 10
Material sphere study · Light and texture
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a photorealistic material and lighting reference still life on a worn wooden table beside a large window. Arrange three cubes in the back row—clear glass, frosted translucent acrylic and matte white—and two rows of spheres showing clear glass, frosted glass, glossy ceramic, rough stone, pale wood, polished chrome, brushed metal, matte black, deep navy velvet and white marble. Surround them with a reflective metal kettle, an iced water goblet, an open reference book, sketch papers, pencil, clear crumpled plastic, crumpled aluminum foil, cream satin, coarse linen and dark velvet. Soft directional daylight from the left, accurate reflections and refraction, visible condensation, tactile surface detail, neutral color grading, landscape 3:2 composition.
Prompt 11
Compact computer · Exploded technical render
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a premium exploded-view technical product rendering of a compact square silver desktop computer against a seamless pale gray-white studio background. Suspend the rounded aluminum top shell above a layered cooling assembly with two black fans, a copper heat spreader and dense heat-sink fins. Center a detailed black logic board with a large high-performance processor and memory modules. Separate the power supply, wireless antennas, speaker chambers, front and rear I/O panels, machined lower chassis, circular ventilation grille, mounting brackets and dozens of small screws in precise assembly order. Use clean orthographic three-quarter perspective, accurate scale, soft shadowing, crisp metal and circuit-board materials, high-end industrial visualization, no brand logo, no watermark, no labels.
Prompt 12
Corner coffee shop · Sketchbook illustration
Reference prompt reconstructed from the supplied image; not the original generation prompt.
Create a highly detailed black ink urban sketch drawn across both pages of an open cream-paper sketchbook on a wooden desk. Depict a lively glass-fronted corner café named “Corner Coffee” at a street intersection, seen from a wide three-quarter angle. Include pedestrians, a cyclist, parked bicycles, street trees, lamps, café tables, patrons inside and outside, a server at the entrance, menu boards and a bilingual street sign reading “幸福路 / Xingfu Rd.” Use fine pen hatching, architectural perspective lines, loose foliage marks and dense observational detail. Show the center fold, paper texture, date notation and a hand holding a black fineliner near the lower edge. Monochrome travel-sketch aesthetic, landscape 3:2 composition, readable principal signage, no watermark.
06 · Workflow
How to use GPT Image 2.5 online
- 01
Choose a workflow
Use text to image for a new idea. For an existing image, switch to reference creation and upload the material you want the model to work from.
- 02
Write the brief
Name the subject, purpose, layout, and light. For an edit, separate what must remain from what should change. Put required wording in quotation marks.
- 03
Check the 3-point default
Confirm GPT Image 2.5 is selected, sign in, and read the current point estimate on the generate button before starting the task.
- 04
Review, download, and refine
Check the image at full size and at its intended display size. Download a useful version; upload it as a reference if another focused revision is needed.
Responsible use
Before you publish an image
Review names, numbers, small text, product geometry, and reference details. Image generation can reinterpret a brief; an edit request does not guarantee that every other pixel remains unchanged.
The official API documents quality tiers including xhigh and max. This page does not currently provide those selectors or a manual 4K output setting. Use the options actually shown in the workspace.
The gallery contains supplied comparison images, not a controlled platform benchmark. The prompts are reconstructed from selected images, not their original generation records. Use images you have permission to upload and review the final asset for its intended use.
Sources
Model information and platform details
Official documentation explains the model family. The platform catalog and pricing page describe this integration. Reviewed September 9, 2026.
07 · FAQ
GPT Image 2.5 FAQ
Can I use GPT Image 2.5 on this page?
Yes. The embedded workspace selects the platform model GPT Image 2.5 by default. Write a prompt or choose reference creation. You need a Nano Banana Pro account and sufficient points to submit a task.
Is GPT Image 2.5 free? How many points does it cost?
The default generation costs 3 Nano Banana Pro points. This page does not offer unlimited free generation. Read the current estimate on the button before submitting; each new generation is a separate task.
How do GPT Image 2.5 and ChatGPT Images 2.5 relate?
ChatGPT Images 2.5 is the OpenAI product experience. GPT Image 2.5 refers to the model family and is also the name of the entry on this platform. Accounts, controls, and billing are separate.
Can I choose Flare or Sunburst here?
The official models are named gpt-image-2.5-flare and gpt-image-2.5-sunburst. This page currently exposes the platform entry gpt-image-2.5, without a separate variant selector. It does not identify that entry as either variant.
Can GPT Image 2.5 edit my existing image?
Use reference creation, upload your image, and explain what to preserve and what to change. For a later revision, upload the approved output again. This is a reference workflow, not an on-canvas brush or mask editor.
Does this page offer 4K, xhigh, max, or Sketch?
Those controls are not currently available in this workspace. Official API quality settings and ChatGPT product features do not automatically appear in a third-party integration. Current choices are shown in the creation panel.
Is GPT Image 2.5 better than GPT Image 2 for my task?
Compare both with the same prompt and references. Inspect typography, subject details, composition, editing changes, and the time and points actually used. The demo comparisons here do not establish a performance ranking.
Are the demo images real generated examples?
The gallery shows supplied before-and-after comparisons. The images are illustrative examples, not independently verified platform tests. The prompt examples are reconstructed from the supplied images and may produce different results. The supplied prompts can be loaded into the workspace for your own tests.
Create next
Open the image workspace Make your next image with GPT Image 2.5
Start with a prompt or reference image. Review the 3-point default, then bring your brief into the workspace.