Twelve GPT Image 2.6 edits, each with the reference images, the prompt that was sent and the image that came back. Copy a prompt, swap in your own references, and run it in the generator.
A circle on the cushion, crosses on the plant and the rug, an arrow for the lamp. GPT Image 2.6 reads each red mark as an instruction and leaves the sofa, the art and the window light untouched.
Follow the red markup: make the circled cushion blue, remove the crossed-out plant and rug, and move the floor lamp to the right. Keep everything else exactly as it is.
The markup example from the GPT Image 2.6 API docs. A single red circle names the vase; the prompt names the color and asks for the circle to disappear from the finished photo.
Recolor the striped vase inside the red circle to solid cobalt blue ceramic. Keep its shape, the dried stems, and the rest of the kitchen unchanged. Remove the red circle from the finished photo.
A marker doodle of a valley at sunrise becomes a photograph. GPT Image 2.6 Sunburst keeps the mountains, the river bend and the tree where the pen put them and invents only the materials and the light.
Turn this drawing into a photorealistic image. Preserve the exact layout, proportions, and perspective. Choose realistic materials and lighting consistent with the sketch intent. Do not add new elements or text.
A ballpoint owl on lined paper comes back as a clean cartoon character with the same proportions, the kind of asset that goes straight into a GIF or a sticker set.
Turn this sketch into a clean cartoon owl character with soft shading and a plain background. Keep the proportions and pose of the drawing, ready to animate.
One photo of a person plus three garment shots. GPT Image 2.6 Sunburst puts the jacket, the top and the boots on her and leaves her face, her pose and the museum behind her alone.
Edit the image to dress the woman using the provided clothing images. Do not change her face, facial features, skin tone, body shape, pose, or identity in any way. Preserve the original background, lighting, and camera angle. Only replace her clothing with the items shown in the reference images.
The street scene from one photo, the dog from another. GPT Image 2.6 Sunburst places the dog beside the woman and matches the lighting and the depth of field so the two never read as a paste-up.
Place the dog from the second image into the setting of image 1, right next to the woman, use the same style of lighting, composition and background. Do not change anything else.
Same knight, same cliff, same castle. Only the armor changes from green and gold to red and gold, and GPT Image 2.6 Sunburst holds the face, the hair and the pose through the edit.
Change the armor from green and gold to red and gold. Keep the face, hair, pose, the cliff and the castle exactly as they are.
A pixel-art game screen is the only reference. GPT Image 2.6 Sunburst borrows its palette, its grid and its chunky outlines to draw a subject that never appears in the source.
Use the same style from the input image and generate a man riding a motorcycle on a white background.
A one-object edit on a real room photo. The white chairs become wood; the camera angle, the floor shadows and everything else in the kitchen stay where they were.
In this room photo, replace ONLY the white chairs with chairs made of wood. Preserve camera angle, room lighting, floor shadows, and surrounding objects. Keep all other aspects of the image unchanged. Photorealistic contact shadows and fabric texture.
A plain bottle photo becomes a sunset billboard with the tagline spelled exactly as written. Flare has been tuned for this kind of poster and product layout in the 2.6 release.
Create a realistic billboard mockup of the shampoo on a highway scene during sunset. Billboard text (EXACT, verbatim, no extra characters): "Fresh and clean". Typography: bold sans-serif, high contrast, centered, clean kerning. Ensure text appears once and is perfectly legible. No watermarks, no logos.
A second-turn edit on the billboard above. One short sentence swaps the sunset for snowfall, and GPT Image 2.6 Sunburst keeps the board, the bottle and the tagline exactly where the first render put them.
Make it look like a winter evening with snowfall.
No reference image, one paper-cut prompt, three settings. Low gives a clean scene fast, medium adds crests and laurels, high fills the stands. The new complexity dial decides how much planning happens before rendering.
Paper-cut collage of a football team lifting a trophy in a packed stadium under a CHAMPIONS banner, layered card stock with visible paper edges. Rendered three times with complexity set to low, medium and high.
Results creators shared on X in the first days of GPT Image 2.6.
Copy any prompt above, add your own reference images and generate with GPT Image 2.6 in seconds.
Winter is coming 🏂 GPT Image 2.6 Prompt: PREMIUM BAROQUE FRAMED ARTWORK - PHOTOREALISTIC GALLERY POSTER Use the uploaded image as the sole visual reference. Create a seamless 3:4 vertical fine-art poster. TOP SECTION (75%): Display the original scene as a beautiful painting Show more
GPT Image 2.6 can create GIFs now :D So I asked Codex to make my character do a heartbeat dance 💗 Here's low, medium, and high mode side by side 👇
【速報】OpenAI、画像モデル「Images 2.6」を公開 ・生成が従来より50%高速に（公式） ・GIFアニメーションの生成に対応 ・APIは速い「Flare」と、高品質で編集の一貫性が高い「Sunburst」 ・本日からChatGPT・Codexの全プランとAPIで提供 検証結果と2.5との比較はリプで↓
your favorite meme account is about to become even more annoying gifs, reference images and faster generation @ChatGPT images 2.6 is giving people more ways to turn random shower thoughts into internet content
The GPT Image 2.6 API gives you more control over image complexity and lets you create GIFs. So as fun short experiment - I asked Flare to create a GIF of a man doing a standing backflip. Compare low, medium and high complexity side by side. 👇 Detailed explanation below Show more
GPT Image 2.6 はモデルの性能向上というより機能追加的な意味合いのほうが強そう？ とくに複雑度(complexity)の設定項目が増えたのが嬉しいポイントな気がする 画像の種類にもよるけど結構影響するっぽい 一枚目が low 二枚目が high （同じ seed じゃないからアレだけど）
