OpenAI released image generation capabilities in ChatGPT in March 2025 as "Images in ChatGPT".
The model excels at generating images that are consitent with the prompt and images used as references, creating images to a particular style or direction.
It is particularly good at rendering text accurately, and following very specific and detailed prompts.
Version 1.5 was released in December 2025 and included more precise editing capabilities and speed improvements.
You can use multiple reference images and can configure quality and size of the generated images.
Qwen Image is a foundation model for image generation and editing from Alibaba.
Qwen is excellent at high-fidelity text rendering across diverse images and languages.
It is also supports a wide range of artistic styles.
Qwen Image is a fast and efficient model.
Nano Banana is excellent at keeping characters consistent and realistic.
It allows multiple reference images to be combined keeping characters and objects consistent.
It is also supports a wide range of artistic styles as well as creating inforgraphics and text rendering.
Nano Banana Pro builds on top of Nano Banana and is more powerful and accurate.
It excells in creating posters, digrams, infographics and product mockups in great details and accuracy.
Pro can access google search results to add context and retrieve real time information.
Can also upscale output resolution and alter aspect ratio.