GPT Images 2.0
Create an agent or open an existing agent in the builder, then follow the steps below. Model and feature availability may depend on your plan.
What it does
OpenAI's second-generation image model became available to agents that generate images.
Before this
An expansion. Agents could already generate images — this replaced the model behind that capability with one that can put legible text inside the picture and follow a layout brief, which is what made the previous generation hard to use commercially.
Why it matters
Noticeably better at text inside images and at following a layout brief — the two things that made the previous generation hard to use commercially.
What changed
This is an image generation model, not a chat model: it powers the image-generation capability you switch on in the Agent Builder, and it is picked from the image model dropdown there rather than the main model selector.
The second generation fixed the two things that kept the first out of commercial work — text rendered inside an image, and following a layout brief instead of approximating it. An agent that produces social graphics, quote cards, or thumbnails can now be trusted with copy in the picture.