OpenAI upgrades ChatGPT's image generator — better detail and text rendering, but still weak outside English

OpenAI released ChatGPT Images 2.0, an improved version of its image-generation tool built into ChatGPT. Testing shows the model creates sharper, more detailed images and can now render readable text within pictures — a major weakness of the previous version.

The upgrade matters most for designers, marketers, and content creators who use ChatGPT to mock up visuals without hiring a designer. You can now ask ChatGPT to generate a poster with legible product names or a mockup with actual readable headlines, instead of getting blurry gibberish text.

Non-English users hit a wall: the model still struggles with languages other than English, meaning requests in Spanish, French, or Mandarin produce lower-quality results. If your team works in multiple languages, you'll need to stick with English prompts or use a different tool.