Two model variants
Flare · Sunburst
Switch variants within the same workflow. Compare results using your own prompts to find the direction that suits your project.
Start with a text prompt or edit a reference image in one familiar workspace. Describe your subject, setting and style, or explain what should change in an existing image. Then choose a model variant and output settings for your project.
Two model variants
Flare · Sunburst
Switch variants within the same workflow. Compare results using your own prompts to find the direction that suits your project.
Two ways to create
Generate · Edit
Create from a written description or add reference images to guide the subject, composition or visual style of an edit.
Three output tiers
1K · 2K · 4K
Choose a resolution tier and aspect ratio for your use case. Exact pixel dimensions depend on the ratio and generation mode.
Tell the model how you plan to use the image so composition, spacing and style have a clear direction. These are starting points to explore, not guarantees of a particular result.
Add a product reference and describe the appearance to preserve, the background, lighting and placement. Check packaging text, product details and brand marks before publishing.
Choose a ratio for a square post, landscape cover or portrait layout. Describe subject placement, background colors and space for text you will add later.
Explore illustrations, posters and scene ideas, or adjust the background, colors and mood of an existing image. Focus on a few changes at a time to compare versions more easily.
Keep the familiar workflow and focus on your description and revisions.
Enter a prompt for text-to-image, or upload references for image editing. Explain what to keep, what to change and how you plan to use the result.
Select Flare or Sunburst below the generation modes, then choose an aspect ratio, a 1K / 2K / 4K tier and the number of images. Enhanced mode requires an explicit ratio.
Review the credit total shown on the generate button before submitting. It depends on the mode, resolution tier, purchase status of your account and image count.
View and download your result. Select a task from workspace history to restore its prompt and settings, then make adjustments before starting a new generation.
Learn about variants, references and output settings before you begin.
Yes. Text-to-image creates an image from your description. Image editing lets you add references and describe what to keep or change. Both workflows support Flare and Sunburst.
They are two selectable GPT Image 2.5 variants with the same workspace controls. Keep your prompt and output settings consistent and compare the results. Both use the same credits for the same generation mode, resolution and account status.
Image editing supports up to 16 reference images. Add images relevant to the subject, composition or style and explain the role of each reference in your prompt. Only upload material you have permission to use.
The aspect ratio determines the shape of the image, while the resolution tier determines its output size level. The workspace offers 12 explicit ratios. Exact dimensions vary by ratio and generation mode; the 4K tier does not mean every image has a 4096-pixel edge. Enhanced mode does not offer Auto ratio.
Start with a clear description or a reference image and explore GPT Image 2.5.
繊細な光と影、リアルな質感で、想像を鮮明に描き出す。
Loading GPT Image 2.5...