OpenAI Beefs Up ChatGPT’s Image Generation Model
The new model creates entire booklets from one prompt and renders text in languages like Chinese and Hindi.
OpenAI has released ChatGPT Images 2.0, a significant upgrade that moves beyond single-image generation. The model can now produce multiple, related images from a single prompt—such as an entire study booklet—and render text directly within images, including in non-English languages like Chinese and Hindi. Leveraging ChatGPT's reasoning capabilities, it can search the internet for recent information, thanks to a December 2025 knowledge cutoff date, making outputs more granular and context-aware. For example, it can generate a detailed infographic with a local weather forecast and suggested activities, complete with accurate landmarks.
Initial tests show marked improvement in text rendering, a previous weakness for AI image models. While English text outputs are cleaner and more complex, performance in other languages is still evolving; the model itself critiqued its own Chinese text output as containing 'semi-gibberish' and malformed characters. The release also introduces greater user control, allowing for customizable aspect ratios from 3:1 wide to 1:3 tall. This global rollout, with a premium tier for paying users, aims to boost engagement by enabling more creative and structured visual content, potentially sparking new social media trends.
- Generates multiple images (e.g., a booklet) from a single prompt, not just one image.
- Renders text within images, supporting non-English languages like Chinese and Hindi with a December 2025 knowledge cutoff.
- Offers customizable aspect ratios from 3:1 to 1:3 and improved, more granular detail in outputs.
Why It Matters
Enables creation of complex, multi-image content for education, marketing, and global audiences, directly within ChatGPT.