ChatGPT offers new image generator

- OpenAI's image generator now uses the powerful 4o model
- Logos, diagrams and fonts are now displayed in more detail
- AI-generated images contain metadata for origin verification

OpenAI has revised its image generator and is now using the powerful 4o model, which replaces DALL-E, reports Heise . This innovation brings more precise implementation of user queries, improved recognition of contexts and optimized display of text within images. There are significant improvements, especially in the areas of logos, diagrams and infographics. In addition, images with transparent backgrounds can now be created, which makes them even more versatile for presentations and marketing purposes.

A significant advance is the ability to create and edit images directly from conversations or from uploaded files. This opens up new application possibilities, such as for individual design adaptations or interactive visualizations. The switch to an integrated model also means that text, audio and image processing work together seamlessly without the need to transfer information between different AI models - a step that improves both efficiency and accuracy.

To ensure transparency, OpenAI adds metadata to all generated images, identifying their origin. At the same time, users are given more freedom in image creation, as content restrictions have been relaxed. However, OpenAI emphasizes that it will closely monitor social developments and ethical issues to ensure the technology is used responsibly.