OpenAIGPT Image 2
GPT-Image-2
GPT Image 2 creates and edits images with flexible dimensions and detailed reference-image input. It remains listed separately from the newer 2.5 and 2.6 releases.
A dated image release remains a separate choice
The documented snapshot is gpt-image-2-2026-04-21. The model supports image generation, editing and Batch processing. Its model card directs cost estimates to the image calculator rather than presenting a fixed charge per finished image.GPT-Image-2 official model documentation
Preserve a known visual workflow
Practical guidance. When an existing asset workflow already uses this release, compare a newer model on the same source images and requested edits. Check subject consistency, text, transparency and output dimensions before changing the model used for an entire collection.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | gpt-image-2gpt-image-2-2026-04-21 |
| Input and output | text, image in; image outGPT-Image-2 official model documentation |
| API routes | Batch (v1/batch); Image generation (v1/images/generations); Image edit (v1/images/edits)GPT-Image-2 official model documentation |