OpenAI's ChatGPT Images 2.0 can now generate text within pictures — a skill that was nearly impossible before

OpenAI released DALL-E 3 Images 2.0, an updated image-generation model that can embed readable text into generated images — something previous versions struggled with. The improvement matters because business users, designers, and marketers often need images with captions, logos, or labeled diagrams.

The model produces text that is legible and contextually accurate to the surrounding image, rather than the garbled or nonsensical letters that older systems produced. This means users no longer need to add text in a separate design tool after generating the image.

For anyone using ChatGPT to create marketing materials, social media graphics, or product mockups, this eliminates a manual post-production step. Teams that rely on AI image generation for quick prototypes can now get closer-to-finished work in a single generation, saving designers hours per week.