According to Bloomberg the new version of ChatGPT Images, announced Thursday, is designed to create and edit images more precisely, and generate them up to four times faster than the previous AI-powered image generation model. The company is also building a new section on the ChatGPT mobile app and website for people to create images, rather than just interact with a chatbot.
OpenAI is increasingly trying to position ChatGPT as a sort of everything app, offering a search engine, a voice assistant, and an image and video generator. But the company is facing increasing pressure from companies like Google, which recently introduced a powerful new AI model called “Gemini 3” and also offers a popular image generator called “Nano Banana”. Elon Musk’s xAI chatbot “Grok” offers similar functionality.
With the latest improvements to the image generator, OpenAI aims to meet the needs of both personal and professional use.
The updated software aims to let users make a variety of edits to uploaded images, such as changing the style from photorealistic to watercolor or giving each person in the photo a new hat – all without losing the detail tracking of previous versions. In particular, OpenAI said the new model can maintain consistent lighting, a person’s appearance, and scene composition across multiple edits and changes. The company’s previous image models sometimes struggled to track such details, especially with multiple edits.
The new model can also generate more and smaller text, the company said, for example to create detailed infographics or multiple paragraphs that users provide in a prompt. It is also better at generating multiple small faces in a single image.
Credits:
Image:
