Back to timeline

ChatGPT Images 2.0 and Thinking Mode

OpenAI launches ChatGPT Images 2.0 with a thinking mode that brings reasoning and tools into image generation.

Model Release

What Happened

On 2026-04-21, OpenAI launched ChatGPT Images 2.0 across ChatGPT plans, alongside a thinking mode for paid users. The official release notes distinguish access to the base image model from access to images with thinking.

Why It Matters

The release connected image generation to a research-and-planning process, allowing visual outputs to draw on live information and multistep reasoning rather than relying only on the initial prompt.

Technical Details

The system card describes live web search, multiple images from one prompt, and reasoning that develops the request before generation. OpenAI reported improvements in world knowledge, instruction following, and dense text rendering. Thinking mode was available through Thinking and Pro selections. These capabilities do not establish a disclosed model architecture; the announcement and system card do not specify parameter counts.