OpenAI said on the 21st (local time) that ChatGPT now includes the new image feature, “ChatGPT Image 2.0.” For developers, its application programming interface will also offer the “GPT Image 2” model.
OpenAI described it as its latest flagship model, with broadly improved image generation and editing. The company said it supports high-quality image inputs and can produce outputs in multiple sizes.
A key focus is stronger text rendering. OpenAI said the tool can more reliably place long sentences and complex visual elements inside images and follow user instructions more accurately. That could make it more useful for posters, ad banners, brochures and presentation visuals where text and design must work together.
OpenAI also said it improved multilingual performance, including for Korean, Japanese, Chinese, Hindi, Bengali and English.
Editing has been refined as well, the company said, allowing users to change specific elements while keeping backgrounds and overall composition intact. “Instruction-following and fine detail have improved overall,” OpenAI said.
OpenAI also highlighted examples of generating sequential scenes while keeping a consistent style and character, a capability it said could help with advertising, promotional materials and series-style image production.
OpenAI said basic image generation is available to ChatGPT users, while more advanced features are centered on paid subscribers. The model is also being applied to the developer API, which is expected to expand integration with outside services.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.