GPT Image 2.5

GPT Image 2.5 is the image model to use when a later video depends on the image holding identity and layout.

First video free, no card

When to use it

Pick it when you are making the image a video will be built on: a portrait, character sheet, first frame, or identity-preserving edit.

Best for

  • portraits and character sheets that a video has to keep
  • the first frame of a clip
  • product in hand
  • edits that must not lose the face

Avoid for

  • bulk throwaway drafts where GPT Image 2.5 Flare is faster

Modes

  • text-to-image
  • image-edit

Prompt notes

  • Describe the subject, age, framing, light, and what the hands do.

  • Use layout instructions such as four poses on a plain background or headroom for a caption.

  • Pass the result to generate_video as first_frame or in refs, depending on whether you need to animate the still or keep the person across clips.

Related use cases

FAQ

What should I use GPT Image 2.5 for?
Use it for portraits, character sheets, first frames, product-in-hand images, and edits where identity matters.
Can I use its output in video generation?
Yes. Use the image as first_frame to animate it or as a reference to keep the person or product consistent.
When should I avoid it?
Use GPT Image 2.5 Flare for faster draft batches when the final series does not depend on a single identity-hold image.

What do you sell?

First video free, no card