GPT Image 2.6 Generator
OpenAI's GPT Image 2.6, in two variants. Sunburst renders heavier, more precise detail. Flare runs faster. Both generate and edit at up to 2K resolution across nine aspect ratios, on getimg.ai.
More than just GPT Image 2.6.
Get access to all top AI image + video models.
Built for precision. Fast when you need it.
See what GPT Image 2.6 brings to every visual format.
Outputs that don't look AI-generated
Across photorealism, cinematic stills, and illustrated styles, GPT Image 2.6 captures the details that make images feel real: grain, ambient light variation, natural framing, consistent texture. Sunburst pushes the finest detail. Flare gets you there faster.
A step up in instruction following
GPT Image 2.6 places objects where you specify, respects spatial relationships, and handles complex multi-element scenes accurately. Small text, UI labels, iconography, and timestamps render as described rather than approximated.
Dense text and multilingual rendering
Strong text-in-image work across the GPT Image family. GPT Image 2.6 handles small type, UI labels, timestamps, and multilingual scripts - Japanese, Korean, Chinese, Hindi, Bengali - with coherent visual layout, not just correct characters.
What GPT Image 2.6 produces best
How to Prompt GPT Image 2.6
GPT Image 2.6 follows a prompt closely, so write it in the order you'd describe a photo: the subject, the exact text, the setting, then light and camera.
Put every word you want in the image inside quotes, and say where it goes: on the glass, on a taped note, across the top of a poster. Write other scripts directly, like Japanese or Hindi, instead of asking for a translation.
Describe the place and the people as specifically as you would to a photographer. Finish with the look: the light, the film stock, the depth of field.
Short prompts work too. Prompt enhancement fills in detail when you start from a single line.
Night street photo of a tiny Tokyo record shop. Gold-leaf lettering on the rain-streaked glass door reads "NIGHT OWL RECORDS", with "ナイトオウル・レコード" in smaller letters underneath. A handwritten paper note is taped below it: "Open till 2am. Jazz upstairs." Inside, an older shopkeeper in a moss-green cardigan flips through vinyl under a green banker's lamp. Cool blue street light outside against the warm interior, reflections of passing umbrellas in the glass. Shot on 35mm film, shallow depth of field.
GPT Image Prompt Templates
Copy a template, fill in the brackets and generate. Each one plays to a different GPT Image strength.
Poster with text
A [style] poster for [event]. Large headline at the top: "[HEADLINE]". Below it: "[date] · [venue]". Small print at the bottom: "[line]". [Two colors] on [paper texture], [layout detail, e.g. bold grid, torn edges].
Keep each piece of text short and in its own quotes. Two or three text blocks render more reliably than a full paragraph.
Edit images with the same precision you get from generation.
GPT Image 2.6 isn't only a generation model - both Sunburst and Flare apply the same instruction-following precision to editing. Attach an existing image and describe what to change: swap a background, modify an object, adjust lighting, or remove an element. Name what should stay the same too, and the model keeps it as it was. For a series of edits on the same image, pick Sunburst: OpenAI says it keeps details more consistent from one edit to the next.
- Swap backgrounds while preserving subject detail and lighting
- Completely change the image style
- Add, modify, or remove objects with accurate spatial placement
- Adjust color, lighting, and style without rebuilding the composition
- Fix imperfections and refine details in existing assets
change the dawing to an ultra-detailed, realistic photo
Up to 2K resolution. Nine aspect ratios. Two variants.
GPT Image 2.6 generates at up to 2K resolution across nine aspect ratios - social, editorial, widescreen, portrait, and standard blog formats - without upscaling or cropping separately. Pick Sunburst for the heaviest detail or Flare for speed; the formats are the same on both.
All generated and edited images include full commercial rights on paid plans. Use them in client work, paid campaigns, merchandise, and social media without additional licensing. Commercial rights are included in every paid plan, not locked behind an enterprise tier.
GPT Image 2.6 vs 2.5 vs 2 vs 1.5
We gave all six GPT Image models in the Create image generator the same prompt: a direct-flash night editorial of a woman in a Seoul alley holding a newspaper with an English headline and a Korean line beneath it, at 16:9 and 1K (GPT Image 1.5 at 3:2, its widest format). All six got the English headline right, and all but GPT Image 1.5 spelled the Korean line correctly.
Explore more AI image models and tools
Frequently Asked Questions
GPT Image 2.6 is the newest generation of OpenAI's GPT Image models, available on getimg.ai in two variants: Sunburst, tuned for heavier, more precise detail, and Flare, tuned for faster generations. Both create images from text prompts and edit existing images from a description, at 1K or 2K resolution across nine aspect ratios, with up to 10 reference images.
OpenAI says GPT Image 2.6 generates images about 50% faster than GPT Image 2.5, and that Sunburst keeps details more consistent from one edit to the next. The formats stay the same: 1K or 2K output, nine aspect ratios and up to 10 reference images. GPT Image 2.5, 2 and 1.5 stay available in the Create image generator if your prompts are already tuned to them.
Start with GPT Image 2.6 Flare to try ideas, then switch to GPT Image 2.6 Sunburst for the final image. Keep GPT Image 2.5 or 2 if your prompts are already tuned to them, and use GPT Image 1.5 for quick square or portrait drafts with English text.
GPT Image is OpenAI's family of image models, the same family behind image generation in ChatGPT. On getimg.ai you pick the exact version, from GPT Image 1.5 to GPT Image 2.6 Sunburst and Flare, and set the aspect ratio and resolution yourself.
You can also attach up to 10 reference images and switch to other image and video models without leaving the Create image generator.
Yes. Attach your photo in the Create image generator, pick a GPT Image model and describe the change: a new background, a different outfit, another lighting setup or a style change. Name what should stay the same too, like the person's face or the product label.
Each edit returns a new image, so your original stays untouched.
Yes. All paid getimg.ai plans include commercial rights. Images generated or edited with GPT Image 2.6 can be used in client work, paid campaigns, social media, and merchandise - included from the Entry plan, not locked behind an enterprise tier. Use is subject to getimg.ai's terms and content policies.
GPT Image 2.6 carries the GPT Image family's strong text rendering. It handles small type, UI labels, icon text, timestamps, and dense typographic compositions. For multilingual content - Japanese, Korean, Chinese, Hindi, Bengali, and Latin scripts - it renders not just correct letterforms but coherent visual layout, from posters to infographics.
On getimg.ai, both GPT Image 2.6 variants support nine aspect ratios: 1:1, 2:3, 3:2, 4:3, 3:4, 4:5, 16:9, 9:16, and 21:9. They generate at up to 2K resolution, covering social, editorial, widescreen, portrait, and standard blog formats - without needing to upscale or crop separately.
Yes. GPT Image 2.6 Sunburst and GPT Image 2.6 Flare are available through the getimg.ai API. You can also use the GPT Image 2.5 Sunburst API, the GPT Image 2.5 Flare API and the GPT Image 2 API to generate and edit images directly in your own tools.
This is separate from the browser experience and is aimed at developers or teams who want to automate image creation. You send prompts or images through the API and get back results that match the quality you see in the app.
It runs on a prepaid, pay-as-you-go basis, so you're not tied to a subscription. Scale usage up or down as you need, with clear pricing and no hidden limits.
It's a practical choice if you're building automated design generation, content pipelines, or anything that needs reliable image output without manual work.
Both are GPT Image 2.6. Sunburst is the heavier variant, tuned for the most precise detail and instruction following - reach for it when the output has to hold up on close inspection. OpenAI also says it keeps details more consistent across a series of edits, so it's the better pick when you edit the same image several times.
Flare is built for speed, so you can generate and iterate faster when turnaround matters more than maximum detail. Both support the same nine aspect ratios, up to 2K resolution, and image editing, and both are selectable in the Create image generator.
OpenAI's GPT Image line has moved through several generations. GPT Image 1 introduced native text-to-image generation and editing. GPT Image 1.5 improved quality and instruction following. GPT Image 2 sharpened instruction following and text rendering and added higher resolution.
GPT Image 2.5 split the line into Sunburst for precision and Flare for speed, and GPT Image 2.6 is the current generation with the same two variants. GPT Image 1 has been removed from getimg.ai, but GPT Image 1.5, GPT Image 2, GPT Image 2.5 and GPT Image 2.6 (Flare or Sunburst) are all available in the Create image generator, so you can pick the version that fits the job.
Precise generation. Precise editing. Your choice of speed.
GPT Image 2.6 is available on getimg.ai now - Sunburst and Flare, up to 2K resolution, nine aspect ratios, and full commercial rights on paid plans. Describe what you want, and the model builds it.