OpenAI’s updated image generator can now pull information from the web
The new 'thinking' model can pull web data and generate up to eight coherent images from one prompt.
OpenAI has rolled out a major update to its AI image generator, launching ChatGPT Images 2.0. This new version is powered by the GPT Image 2 model and introduces groundbreaking 'thinking capabilities' for Plus, Pro, Business, and Enterprise subscribers. When this mode is enabled, the tool can actively search the web to gather information and context before generating images. This allows it to 'reason through the structure of an image' and create sophisticated visual explainers based on uploaded files. The most significant new feature is the ability to generate a coherent series of up to eight images from a single prompt, maintaining consistent characters, objects, and styles across all outputs.
Beyond its new reasoning engine, ChatGPT Images 2.0 brings substantial quality-of-life and technical improvements for all users. Image resolution has been increased to 2K, and the tool now supports a wider range of aspect ratios, from a cinematic 3:1 to a tall 1:3. It shows 'significant gains' in rendering text within images for non-Latin scripts like Japanese, Korean, Chinese, Hindi, and Bengali. OpenAI is positioning this update as a direct response to increasing competition from tools like Google's Nano Banana Pro and Microsoft's MAI-Image-2, aiming to solidify its lead in creative AI applications for professionals and businesses.
- New 'thinking' model can search the web for context before generating images, a first for OpenAI's image tools.
- Generates up to eight stylistically consistent images from one prompt, ideal for comics, social campaigns, or design mockups.
- Outputs 2K resolution images, supports wide aspect ratios, and renders text significantly better in Asian languages.
Why It Matters
Moves AI image generation from single illustrations to coherent visual storytelling, unlocking new workflows for creators and marketers.