GPT Image 1 I2I

GPT Image 1 image to image is OpenAI's multimodal edit mode that brings reasoning, world knowledge, and visual generation into one unified system. It takes an input image plus a text instruction and applies targeted changes such as object replacement, background swaps, lighting adjustments, or localized restyles while preserving untouched regions and subject identity. Because it shares architecture with GPT Image 1 text to image, edits benefit from the same strong prompt adherence and embedded text rendering, which makes it practical for revising product shots, iterating on ad creative, or cleaning up rough drafts. Output sizes match the text to image variant at 1024 by 1024, 1536 by 1024, and 1024 by 1536, with 4K available through upscaling when higher fidelity is required for print or large format use.

April 23, 2025

OpenAI

Image to Image (I2I)