OpenAI's Image Generator Gains Web Search, Boosting Creative Control for Businesses

OpenAI has released a substantial upgrade to its ChatGPT image generator, equipping it with the ability to search the web. This new feature, part of the ChatGPT Images 2.0 update, aims to provide more nuanced and detailed visual outputs by allowing the model to gather external information before creating an image.

The core advancement is a new 'thinking' mode, available to paid-tier subscribers. When activated, the model can research topics online, analyze uploaded files to build visual explainers, and plan an image's composition before generating it. This process supports the creation of multiple images—up to eight at once—that maintain consistent characters, objects, and styles across a series. For marketing teams or product designers, this could streamline the production of social media campaigns, storyboard sequences, or architectural visualizations.

All users gain from broader enhancements. The tool now produces images with up to 2K resolution and a wider range of aspect ratios, from panoramic 3:1 to vertical 1:3 formats. It shows improved handling of specific visual styles, including pixel art and cinematic stills. Notably, OpenAI reports major progress in rendering accurate text within images for several non-Latin scripts, including Japanese, Korean, and Hindi.

The update arrives as competition intensifies. Google and Microsoft have recently launched their own advanced image generation models. By integrating web search and multi-image consistency, OpenAI is betting that greater creative precision and workflow utility will distinguish its offering for professional users.

Source: The Verge