Washington, April 22 (QNA) - OpenAI, a leading AI research company, announced a new update to its image generation mechanism in ChatGPT, adding what the company describes as "thinking capabilities," enabling online search and the creation of a set of related images based on just one command.

The update is based on the new GPT Image 2 model, which enhances the accuracy of instruction execution, preserves user-defined details, and improves the generation and display of text within images. It also enables the activation of Thinking mode to analyze image structure and utilize online information, as well as create visual annotations and infographics based on user-uploaded files.

In the same context, the new version allows the creation of up to eight images at once, while maintaining the same elements such as characters, objects and style across different scenes.

The company noted that the model also supports the production of Japanese manga comic pages, designs for social media platforms, and more.

The update includes general improvements for all ChatGPT users, including increased accuracy in capturing basic image features, support for multiple styles such as pixel art, manga, and cinematic shots, in addition to providing resolutions up to 2K, with various aspect ratios.

Regarding text within images, OpenAI explained that the new model provides significant progress in generating text in non-Latin languages, with particular improvement in displaying English text.

The company indicated that the ChatGPT Images 2.0 model is now available to all ChatGPT and Codex coding app, while the new 'thinking' model are reserved for subscribers of paid plans, such as ChatGPT Plus, ChatGPT Pro, and ChatGPT Business. (QNA)