OpenAI launches ChatGPT Images 2.0 with improved text and multi-image output

In short: OpenAI has released ChatGPT Images 2.0, a new image-making model that can create multiple images at once and is better at putting readable text into images.

What happened

OpenAI launched a new image generation model on Tuesday called ChatGPT Images 2.0. Image generation means the system creates pictures from a written request, like asking an artist to draw something based on your description.

OpenAI says the model can produce more than one image from a single prompt, like pages of a small study booklet instead of just one picture. It can also add text inside images, including in languages such as Chinese and Hindi.

The update is available globally for ChatGPT and Codex users. A more powerful version is available for paying subscribers.

Why it matters

People often use AI image tools for posters, menus, class handouts, and social posts, and these tasks usually need both pictures and readable words. Wired’s early testing found Images 2.0 does a better job with English text than older image tools, which often produced misspelled or broken-looking letters.

However, the same testing also suggests the tool may still struggle with languages other than English. In one example, a Chinese-style collage included text that looked real but was partly nonsense, and ChatGPT itself said much of it did not translate cleanly. For everyday users, that means you may still need to double-check any text, especially if it is not in English.

Source: Wired