Create posters, product visuals, and photo edits with GPT Image 2.5 Flare, OpenAI’s speed-focused image model. Start with a text brief or upload reference photos, then generate directly on this page.
See a photo restyle, a two-image makeup edit, and a text-led information graphic. Copy a prompt to try in the generator above, or open the linked case to view its saved setup. Use your own reference photos for edits.
Rework a photo’s setting and lighting from an existing image. This case pairs a tabletop reference with a request for a brighter background and softer shadows.
Use the uploaded photo as the reference. Explore a brighter studio background and softer shadows while focusing the composition on the main product.
One portrait supplies the makeup look; another supplies the person. Both inputs are shown with the finished edit so you can see how the reference roles are used.
Transfer the makeup style from Image 1 to the person in Image 2. Only change the makeup on the person in Image 2, including eyeshadow, eyeliner, lashes, blush, contour, lip color, and overall makeup finish. Keep the person’s facial features, identity, hairstyle, clothing, pose, lighting, background, and photo style unchanged.
Turn a written brief into a visual sequence. This text-only example uses numbered stages, arrows, and specified labels to explain a simple rainwater journey.
Create a finished vertical 4:5 editorial infographic titled FROM RAIN TO TAP. Warm ivory background, navy typography, restrained teal accents. Arrange exactly four numbered steps in a clear top-to-bottom sequence connected by arrows. Exact labels: 1 COLLECT; 2 FILTER; 3 DISINFECT; 4 DELIVER. Each step has one simple flat illustration: a reservoir under a rain cloud, a filter tank, a treatment chamber, and a household tap filling a glass. Include exactly this footer: A SIMPLE WATER JOURNEY. Use only the specified text, with exact spelling and no extra words. Large readable lettering, generous margins, consistent icon style, professional educational design. Do not repeat any step or add a fifth step. No logos, watermark, photorealistic background, or decorative illegible writing.
GPT Image 2.5 Flare turns text and image inputs into images. OpenAI positions it for fast everyday creation, while Sunburst focuses on demanding quality and editing precision. Use Flare to develop a new visual or rework an existing photo.
Flare is the speed-oriented option in GPT Image 2.5. It is a starting point for exploring visual ideas and alternatives when frequent iteration matters. Actual generation time depends on the request and settings.
Create posters, labeled visuals, and information graphics that combine words with imagery. The information-graphic example shows a text-led brief organized into numbered stages and arrows.
Use an existing person, product, or scene as the starting point for a change. References can supply different ingredients: the makeup example uses one portrait for the look and another for the person.
Choose the starting point for your task. Both options use the Flare generator at the top of this page.
No source image is required. Describe the subject, setting, style, and composition—for example, a product concept or a poster. Include any words that should appear in the design.
Upload your own photos and explain the change you want. For several references, identify what each one supplies, such as a person, makeup look, outfit, or background.
Start with the image you want. For edits, explain what should change and what should stay. These six real examples show how that direction appears in a complete prompt. Copy one, return to the generator, and adapt it to your subject and references.
For a simple edit, describe the change directly. The dog example asks for an astronaut suit, while the uploaded photo provides the starting subject.
Dress the puppy in the picture in an astronaut Space suit
State the requested change and retained details separately. The product example changes the background while naming the shape, label, cap, and position to keep.
Replace the existing background with a clean warm-white photography studio. Keep the product’s shape, color, proportions, label, cap, reflections, and position unchanged. Add a soft natural shadow directly beneath the product so it remains grounded on the surface. Do not alter the product design, camera angle, crop, or packaging text.
Identify which image supplies each ingredient. This fashion prompt assigns separate references to the person, clothing, and background.
The model in Figure 2 is wearing the clothes from Figure 1 and standing against the background of Figure 3.
Describe where the artwork should leave space for copy. The plant-market example places leaf illustrations around a clear central area; it does not supply exact event wording.
A minimal poster concept for a weekend plant market, illustrated green leaves, cream background, a clear central space for the event title and details.
Describe how the product belongs in a scene. The pet-bed example combines the product reference with a cat, rug, sofa, and soft daylight for a home setting.
Based on the pet product images I uploaded, generate an image depicting a European/American family pet lifestyle. Please strictly preserve the original appearance, color, material, size proportions, and functional details of the product; do not alter the product structure. The scene should depict a clean, natural Ragdoll cat using the product in a light-colored living room, surrounded by a beige rug, light gray sofa, wooden side table, and soft sunlight. The product should be clearly visible, and the pet's expression should be natural, avoiding excessive anthropomorphism, cartoonish elements, fake smiles, or unnecessary limbs. The overall style should be warm, believable, and realistic, like a high-quality advertisement from a European/American pet brand's official website. Do not include Chinese characters, promotional text, watermarks, or incorrect logos.
Explain which objects overlap to create depth. The phone-poster prompt separates a foreground hand and phone, a character projecting past the screen, and pavement behind them.
Create a 3:4 vertical POV surreal "pop-out" mobile poster. The subject is Sun Wukong, retaining the classic animated look from the reference image: monkey face, golden headband, yellow top, light blue scarf, red trousers, and black boots. Maintain the original pose: one arm raised and bent beside the head, the other hand at the waist, one leg bent and lifted, and the other extending downward toward the bottom of the phone screen. Do not change this to a standard standing pose or a pose involving the swinging of the golden staff. Use a first-person, wide-angle, top-down perspective. A hand holds a black iPhone XR from the bottom-left of the frame; the phone is positioned in the lower half, slightly tilted, with the device fully visible and the hand-to-phone scale realistic. The character's head and shoulders are close to the lens and significantly enlarged, while the lower body diminishes in size with distance, creating a strong forced perspective. The phone screen acts as a spatial portal. Sun Wukong's head, shoulders, chest, and raised arm clearly protrude from the screen, obscuring the top edge and parts of the sides, while the legs extend into the screen's interior. The entire body is continuous and intact, creating a clear 3D pop-out illusion through frame occlusion, contact shadows, and depth within the screen. It is neither a flat image on the screen nor a separate figure floating above the phone. The background outside the phone is a realistic city sidewalk with gray concrete and natural daylight with soft shadows; the interior of the phone screen retains a bright blue sky with clouds, creating a contrast between the virtual world and the real ground. The hand and phone have a realistic photographic texture, while Sun Wukong retains his classic animated silhouette and color scheme, enhanced with three-dimensional shading and a sense of spatial depth. At the top of the image, include large, rounded white text reading "DESIGN" and "HAPPY" on two separate lines. The text is positioned behind the character and naturally partially obscured by the head and shoulders, without blocking the face, hand gestures, or the phone itself. The composition is clean and lively, possessing the strong visual impact of a social media poster. Feature only Sun Wukong, the hand holding the phone, and the phone itself; do not add other characters, interface buttons, or extra text. Avoid duplicated limbs, deformed fingers, body fragmentation, incorrect perspective, and conflicting lighting directions.
Use the generator at the top of this page. You can begin with your own brief or copy a prompt from an example.
Write a brief for a new image, or upload photos for an edit. If you copied an example prompt, paste it into the description field and replace any reference assignments with your own.
Select the aspect ratio, image size, quality, and number of images. The generator displays the points required for the selected options before you submit.
Click Generate, then open the result to review the image and any required text. Adjust the prompt for another attempt, or upload the result as a reference for a further edit.
Use the model roles below to choose a starting point. OpenAI provides the positioning; size and quality choices reflect the current WeShop form.
| Dimension | GPT Image 2 | GPT Image 2.5 Flare | GPT Image 2.5 Sunburst |
|---|---|---|---|
| Starting priority | An established GPT Image 2 workflow | Fast everyday generation and visual iteration | Demanding image quality or editing precision |
| Starting inputs | Text and image references | Text and image references | Text and image references |
| When to consider it | Your existing prompts already deliver the result you need | You want to explore new images or everyday reference edits with speed as a priority | A detailed brief or precise edit does not meet your requirements with the other choice |
| Image sizes in WeShop | 1K, 2K, 4K | 1K, 2K, 4K | 1K, 2K, 4K |
| Quality choices in WeShop | Low, Medium, High | Low, Medium, High, XHigh, Max | Low, Medium, High, XHigh, Max |
OpenAI recommends Flare as a starting point for fast everyday generation. Use your actual prompt and references to decide whether the result meets your needs.
If an image or edit falls short of your requirements, compare it with Sunburst using the same inputs and output settings. Choose by the result you need; a quality label alone does not establish equal output quality across models.
Open a related model page to review its capabilities, examples, and current generation options.
Practical answers before you start.
It is OpenAI’s speed-focused GPT Image 2.5 model for creating and editing images from text and image inputs.
You can create posters and information graphics from text, or use reference photos for tasks such as changing a setting, restyling a subject, and combining visual ingredients. The examples on this page show actual Flare requests and results.
No. Use text alone for a new image. Upload photos when the request needs an existing person, product, scene, or style to guide it.
OpenAI positions Sunburst for demanding quality requirements and editing precision. If Flare does not meet your brief, compare the same prompt and references with Sunburst before choosing a model for that task.
Click Copy Prompt, return to the generator above, and paste it into the description field. For edits, upload your own reference photos and update the image roles. Use This Prompt opens the linked case and its saved setup; it does not automatically fill this page’s form.
The current WeShop form lists 1K, 2K, and 4K image sizes, with Low, Medium, High, XHigh, and Max quality. Select the aspect ratio and image count in the form, and check the displayed points before generating.
Check spelling, labels, composition, and any person or product details that matter to the task. If detail falls short, try a more specific prompt or compare a higher quality setting; a higher setting does not guarantee a better result for every request.