GPT Image 2.5 Generator and Editor
What you can create with GPT Image 2.5
These feature illustrations explore three prompt ideas: exact lettering, a new visual medium, and a focused reference edit.
Build a poster around exact lettering
Specify the word, type style, placement, and subject. Review generated lettering before using the finished image.
Reimagine a subject in a new medium
Describe details such as cut edges, paper grain, and palette. These are creative directions in a prompt, not selectable style presets.
Guide a focused edit with a reference
Say what should stay and what should change. Image to Image generates a new image, so it cannot guarantee that every untouched pixel will remain identical.
How to use GPT Image 2.5 on Seavid
Choose a generation or editing mode, describe the image you need, and set the output before you generate.
Choose a mode and variant
Use Text to Image for a new image, or Image to Image when you have reference images. Choose Flare for general image work or Sunburst for detailed prompts.
Write a specific prompt
Describe the subject, framing, lighting, and intended use. For an edit, state what should change and what must stay; Image to Image does not lock untouched pixels.
Set the output and generate
Choose 1K, 2K, or 4K and one of the available aspect ratios. Each image costs 10, 16, or 25 Seavid credits respectively.
Explore More AI Creation
Continue with related workflows, models, tools, and effects.
GPT Image 2.5 controls on Seavid
Flare and Sunburst
Choose Flare for general image generation. Sunburst handles detailed prompts and can take longer to generate.
Text-to-image generation
Describe a subject, scene, or layout to create a new image without uploading a source image.
Reference-image editing
Upload up to 16 images in Image to Image mode and describe the revision you want to make.
Three output resolutions
Choose 1K, 2K, or 4K. The Seavid credit cost is 10 credits at 1K, 16 at 2K, and 25 at 4K per image.
Nine aspect-ratio choices plus Auto
Use Auto, 1:1, 3:2, 2:3, 16:9, 9:16, 4:3, 3:4, or 21:9 for the frame.
Common image formats
Reference uploads accept JPEG, PNG, and WebP files up to 30 MB each.
GPT Image 2.5 FAQ
Learn how to choose a variant, edit with references, and check output limits and credit costs.