GPT Image 3 Agent

GPT Image 3 is a browser-based AI image workspace for generation, reference-based editing, and conversational refinement. Users can start from a text prompt or upload one or more reference images, describe the desired scene, layout, format, or correction, and continue editing the chosen result in the same conversation. The workspace keeps chats and generated assets connected, supports @ mentions for targeting an exact source or result, and lets users switch between supported image models and provider routes. It is suited to product scenes, campaign variations, brand graphics, ecommerce visuals, social assets, and other creator workflows.