GPT Image 2 (ChatGPT Images 2.0)

OpenAI's first image model that reasons before it draws. In Thinking mode it can search the web, plan a layout and check its own output, and it generates up to eight images from one prompt with the same characters and objects carried across the set. It added 2K output, aspect ratios from 3:1 to 1:3 and much stronger text and non-Latin script rendering. Instant mode brought the quality gains to every ChatGPT user, free tier included.

For AI filmmakers the eight-image batch was the headline: a set of consistent frames from a single prompt is a storyboard, or the start frames for a sequence. It arrived a month after OpenAI announced the end of Sora, and three weeks before it retired DALL·E 2 and 3 on May 12, which left image generation as OpenAI's way into AI film. GPT Image 2.5 followed in September.