OpenAI's newest image model, released September 8, 2026
Turn a prompt into an image with GPT Image 2.5. Choose Flare for fast drafts or Sunburst for detailed work, then set the aspect ratio, resolution, and background before you generate.
A dreamlike desert library carved into warm sandstone cliffs, one arched reading room glowing with amber light, a lone traveler on the steps, windblown sand, cinematic dusk, intricate architectural detail, painterly photorealism, vertical composition.
A premium studio product photograph of a translucent cobalt blue glass perfume bottle on rough pale travertine, crisp water reflections, soft angled sunlight, minimalist editorial composition, tactile surface detail, vertical composition.
An elegant whimsical illustration of a fox astronomer in a dark green velvet coat studying a brass star map inside a moonlit observatory, a huge telescope behind, warm lamplight and deep indigo night, rich storybook detail, vertical composition.
A rain-soaked Tokyo side street at blue hour, one small ramen shop glowing beneath paper lanterns, reflections on wet pavement, a person with a transparent umbrella walking away, atmospheric urban photography, vertical composition.
A playful high-end editorial still life of three bright citrus fruits, a folded lilac linen napkin, a polished silver spoon and a small blue ceramic vase on a peach tabletop, hard afternoon shadows, bold composition, fine photographic detail, vertical composition.
An otherworldly alpine valley with a narrow stone bridge crossing turquoise glacial water toward a floating garden island, misty distant mountains, tiny wildflowers in the foreground, luminous early morning sky, cinematic matte painting, vertical composition.
A dreamlike desert library carved into warm sandstone cliffs, one arched reading room glowing with amber light, a lone traveler on the steps, windblown sand, cinematic dusk, intricate architectural detail, painterly photorealism, vertical composition.
A premium studio product photograph of a translucent cobalt blue glass perfume bottle on rough pale travertine, crisp water reflections, soft angled sunlight, minimalist editorial composition, tactile surface detail, vertical composition.
An elegant whimsical illustration of a fox astronomer in a dark green velvet coat studying a brass star map inside a moonlit observatory, a huge telescope behind, warm lamplight and deep indigo night, rich storybook detail, vertical composition.
A rain-soaked Tokyo side street at blue hour, one small ramen shop glowing beneath paper lanterns, reflections on wet pavement, a person with a transparent umbrella walking away, atmospheric urban photography, vertical composition.
A playful high-end editorial still life of three bright citrus fruits, a folded lilac linen napkin, a polished silver spoon and a small blue ceramic vase on a peach tabletop, hard afternoon shadows, bold composition, fine photographic detail, vertical composition.
An otherworldly alpine valley with a narrow stone bridge crossing turquoise glacial water toward a floating garden island, misty distant mountains, tiny wildflowers in the foreground, luminous early morning sky, cinematic matte painting, vertical composition.
SHOWCASE
Every image in this gallery was made with GPT Image 2.5 in this generator. Hover over an image, or tap it on touch screens, to see its full prompt and settings. Use the prompt to start your own image.
Create a striking macro studio photograph of a single large sculptural flower made of liquid chrome and translucent iridescent glass, its petals bending like blown metal with electric cyan, violet and hot magenta reflections. Deep midnight blue velvet folds beneath it, dramatic raking light, crisp reflective material, high-end art direction, a bold unforgettable silhouette. Vertical composition. No text, typography, logo, watermark, frame or collage.
Fashion editorial portrait of a young woman in a sculptural plum silk coat on a rain-washed rooftop, wind lifting one sleeve and a distant city softly blurred behind her. Vertical crop, cinematic natural light, detailed fabric and skin texture, confident candid gaze. No words, logos, or watermark.
Premium ceramic tea vessel with matte celadon glaze and a delicate branch of white blossoms on a dark walnut table, in a quiet contemporary tea studio. Square composition, soft window light, tactile clay texture, editorial product photography. No words, logos, or watermark.
A retro adventure comic scene of a brave botanist discovering enormous bioluminescent mushrooms in a cavern. Expressive ink lines, bold coral and cobalt halftones, clear visual storytelling, dramatic shadows, wide canvas. No speech bubbles, words, logos, or watermark.
A quiet ink wash landscape of layered mountain ridges, a tiny red footbridge over a misty river, and one pine tree leaning into the composition. Panoramic view on warm rice paper, expressive black brushwork, restrained vermilion accent, visible watercolor bleed. No words, seals, logos, or watermark.
A surreal greenhouse suspended among soft clouds, glass walls enclosing giant ferns and a winding stone path, with tiny birds crossing the foreground. Vertical cinematic matte painting, luminous dawn, moss green and pale gold tones, believable architecture and atmospheric depth. No words, logos, or watermark.
A breathtaking underwater ballroom inside a vast Art Nouveau glass conservatory beneath the sea. Human dancers in flowing coral-pink silk move across a mirrored floor while translucent jellyfish glow like chandeliers overhead, and schools of fish drift beyond enormous arched windows. Wide cinematic fantasy artwork, opulent yet believable architecture, luminous turquoise water against warm rose gold light, deep layered perspective, richly detailed costumes. No words, logos, watermark, border or collage.
A pair of sculptural running shoes in pale sandstone mesh on a monolithic terracotta plinth, with fine dust in the air and a warm curved studio backdrop. Square premium editorial product photograph, directional side lighting, precise textile and rubber detail, spacious composition. No brand marks, words, logos, or watermark.
A whimsical paper-cut city made of stacked pastel buildings, small arched bridges, and rooftop gardens, seen from a slightly elevated angle. Vertical handcrafted layered paper texture with crisp edges, soft shadows, and a refined coral, mint, and cream palette. No text, logos, or watermark.
An arresting graphic editorial portrait of an anonymous musician in three-quarter profile, rendered as high-contrast black ink and halftone on a saturated golden-yellow field. Wild windswept hair and energetic angular brush strokes burst past the silhouette; one small cobalt blue accent, tactile screen-print grain, dynamic asymmetrical poster composition, sophisticated contemporary art direction. Portrait orientation. No lettering, text, logo, watermark, border or collage.
A monumental kinetic sculpture of suspended brass rings, emerald and cobalt glass spheres, and tiny orbiting mirrored weights in a soaring white museum hall. Tall vertical architectural photograph, elegant asymmetrical balance, precise reflections and long late-afternoon shadows across pale stone, airy editorial mood, exquisite material detail, a human visitor far below for scale. No text, labels, logos, watermark, frame or collage.
A contemporary dancer in a sweeping crimson silk dress suspended at the peak of a powerful leap, the fabric exploding into long sculptural ribbons across a charcoal concrete theater. Horizontal cinematic dance photograph, entire athletic pose visible, dramatic directional spotlight, frozen motion and sharp fabric detail, intense red against near-black negative space, striking editorial campaign image. No text, logos, watermark, frame or collage.
HOW IT WORKS
You don't need an OpenAI API key or any code. Everything happens in the generator at the top of this page, and the whole process takes three steps.
Write what you want to see: the subject, the setting, the style, the light, and any words that should appear in the image. To edit a picture instead, switch to Image to Image, upload a reference image, and describe only what should change. GPT Image 2.5 is built to change the part you name and keep the rest, so short, specific edit instructions work best.
Choose Flare for quick drafts and variations, or Sunburst for detailed final work. Then pick the aspect ratio (Auto or one of 8 presets), the output resolution (1K, 2K, or 4K), and the background (automatic, opaque, or transparent).
Preview the result or download it. Your images are saved to your dashboard so you can find them later. Download a copy for your own backup.
GPT Image 2.5 is OpenAI's latest image generation model, released on September 8, 2026. Inside ChatGPT it appears as ChatGPT Images 2.5, which is why “ChatGPT Image 2.5” and “GPT Image 2.5” point to the same model family. OpenAI rolled it out to ChatGPT, ChatGPT Work, and Codex across all tiers, and released it to developers as two API models: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.
There is no single model called “gpt-image-2.5”. The API model IDs are gpt-image-2.5-flare and gpt-image-2.5-sunburst, and both are also listed in Microsoft Foundry. This generator gives you both models in one place, with no code. OpenAI describes the release as bringing sharper details, more precise editing, and faster generation.
MODEL GUIDE
Flare makes it easy to explore quickly. Sunburst gives demanding visuals more attention to detail. Both work with the same controls in this generator, so you can switch models without rewriting your prompt.
Flare is the small model, optimized for speed. It is the right choice for drafts, variations, and quick iterations when you want to see many ideas fast.
Sunburst is the base model, optimized for quality. It takes longer to generate but holds intricate detail better, which makes it the pick for final assets, dense layouts, and edits where small elements need to survive.
Both models support improved editing and subject preservation. In this generator, both offer 8 aspect ratios plus Auto, 1K / 2K / 4K output, and three background modes. Their published API rate is also the same: $30 per 1M image output tokens. Actual cost per image and wait time depend on your settings and token use.
OpenAI's launch post says Flare delivers higher quality than GPT Image 2 at 50% lower latency, while OpenAI's developer guide describes Flare's quality as comparable to GPT Image 2. We show the more conservative figure above, so you can pick a model based on what you will actually see.
Model facts: OpenAI prompting guide · OpenAI image API guide
WHAT'S NEW
GPT Image 2.5 is the next step after GPT Image 2. Here is what changed, based on OpenAI's launch notes and developer documentation.
| GPT Image 2 | GPT Image 2.5 | |
|---|---|---|
| Released | April 2026 | September 8, 2026 |
| API models | gpt-image-2 | gpt-image-2.5-flare and gpt-image-2.5-sunburst |
| Speed | Baseline | Up to 50% lower latency |
| Image quality | Baseline | Flare: comparable · Sunburst: higher |
| Targeted edits | Supported | Better at changing only what you name |
| Consistency across edits | Supported | Steadier details over multiple rounds |
OpenAI reports up to 50% lower latency than Images 2.0, so you can try more directions in one session.
Change a product, a background, or a line of copy without rebuilding the whole image.
Key details hold steadier over several rounds of edits, so a series of revisions still looks like one piece of work.
The model handles more complex layouts, follows a requested style more closely, gets real-world details right more often, and supports transparent backgrounds.
FEATURES
Creation controls, made clear. Every setting maps to a documented GPT Image 2.5 option, so this AI GPT image generator sends the model exactly what you choose.
Export at 1K, 2K, or 4K. GPT Image 2.5 supports up to 3,840 pixels on the long edge, which is enough for print layouts and large screens.
Pick a square, portrait, or widescreen preset, or let Auto decide. Choose from 1:1, 3:2, 2:3, 16:9, 9:16, 4:3, 3:4, 21:9.
Choose transparent, opaque, or automatic backgrounds. Transparent output is ideal for product cutouts, stickers, and logos you will place on your own designs.
Upload a reference image and describe the change. Use it to swap a background, restyle a photo, or update a product while keeping the rest of the picture intact. Add up to 16 reference images.
Posters, labels, signs, and interface mockups need words that read correctly. Put the exact wording in quotation marks in your prompt, and use Sunburst with a higher resolution when the layout has small print.
Write a full creative brief instead of a handful of keywords. Detailed prompts help GPT Image 2.5 keep composition, style, and small details on target.
USE CASES
A starting point for visual work across product, editorial, and campaign projects.
Place a product in a clean studio scene, then change the label or background while the lighting stays the same.
Make post graphics and carousel sets in the exact aspect ratio each platform needs, from square feeds to vertical stories.
Build text-led layouts where the headline and the small print both need to be legible, then check every label before you publish.
Explore visual narratives for articles and stories in a consistent style.
Place an idea into a believable presentation scene, or sketch interface concepts that keep a clear visual hierarchy.
Create a series of on-brand campaign assets that share the same visual direction instead of drifting from image to image.
WHY CREATE HERE
Create, explore, and keep your GPT Image 2.5 images in one place, with a workflow built around your ideas.
01 / 04
Start with a written prompt, or add a reference image when you want to build on something you already have. Both ways to create live in the same generator, so you can switch between text to image and image to image without leaving the page. You can explore either approach while keeping your creative direction in view.
Create with GPT Image 2.5
02 / 04
Decide how the image should come out before you generate. Pick Flare or Sunburst, choose the aspect ratio, set the resolution to 1K, 2K, or 4K, and make the background transparent when you need a cutout. With these controls in one place, you can prepare a direction that fits your intended use before creating the image.
Create with GPT Image 2.5
03 / 04
Not sure where to begin? Browse the example images, pick one that matches your idea, and send its prompt to the generator with one click. Change a few words, keep what works, and make it your own. The examples give you a concrete starting point without locking you into a template or a fixed style.
Create with GPT Image 2.5
04 / 04
Every image you create is saved to your dashboard, so you can find it later and download it again. Your generation history keeps finished images together after you leave the generator. Download a copy for your own backup whenever you need one. The dashboard gives you a place to revisit completed work without relying on a temporary result screen.
Create with GPT Image 2.5PLANS & CREDITS
Start with one free image. Choose a monthly credit plan or buy credits when you need them.
The same credits each month. Yearly plans are billed upfront.
Try the generator on your own terms.
One free generation per account, for life.
1 free generation
For occasional image projects.
$238.80 billed yearly
Save $96 vs monthly billing
720 credits / month
Up to 360 low-quality images without references
About 36 credits per $1
More room for regular creative work.
$358.80 billed yearly
Save $156 vs monthly billing
2,400 credits / month
Up to 1,200 low-quality images without references
About 80 credits per $1
For a higher volume of images.
$958.80 billed yearly
Save $420 vs monthly billing
7,200 credits / month
Up to 3,600 low-quality images without references
About 90 credits per $1
ONE-TIME CREDITS
Buy a pack once and use it over 12 months. Packs work on every plan, including Free.
A little extra for your next idea.
210 credits
Up to 105 low-quality images without references
About 21 credits per $1
More credits for a growing project.
800 credits
Up to 400 low-quality images without references
About 27 credits per $1
A larger reserve for ongoing work.
2,400 credits
Up to 1,200 low-quality images without references
About 30 credits per $1
Low, medium, and high quality cost 2, 4, and 16 credits per image. Each reference image adds 2 credits.
Prices shown include tax.
FAQS
GPT Image 2.5 is OpenAI's image generation model released on September 8, 2026. It creates images from text and edits existing images, and it comes in two versions: Flare, built for speed, and Sunburst, built for detail. You can use both in this generator.
Flare is the smaller, faster model and is best for drafts and quick iterations. Sunburst is the base model; it takes longer but produces higher image quality and holds fine detail better. Both use the same controls here and share the same published API rate.
Yes. Sign in and your account gets 1 free image.
OpenAI released GPT Image 2.5, also called ChatGPT Images 2.5, on September 8, 2026. The API models GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst launched the same day.
No. GPT Image 2.5 Generator is an independent tool that gives you access to OpenAI's GPT Image 2.5 models. It is not affiliated with or endorsed by OpenAI. The model itself is developed by OpenAI.
Yes. OpenAI calls the ChatGPT experience “ChatGPT Images 2.5”, and the underlying models are GPT-Image-2.5 Flare and Sunburst. People search for both names, but they refer to the same release.
You can choose 1K, 2K, or 4K output and 8 aspect ratios plus Auto. GPT Image 2.5 supports up to 3,840 pixels on the long edge and aspect ratios from 1:3 to 3:1.
Yes. Switch to Image to Image, upload a reference image, and describe what should change. GPT Image 2.5 is designed to edit the part you name while keeping the rest of the image consistent. You can upload up to 16 reference images.
Yes. You can download every image you create. Images are also saved to your dashboard so you can find them later. Download a copy for your own backup.
Write one clear sentence, pick Flare or Sunburst, and see your first image in moments.
Create with GPT Image 2.5
