Google is preparing to launch Nano Banana 2, a new image creation artificial intelligence (AI) model in the Gemini 2.5 Flash ecosystem, which is considered a great step forward in the ability to understand and display images.
According to TechRadar, Nano Banana 2 will be a comprehensive upgrade compared to the first version, with a wider landscape window, supporting many frame rates (1:1, 3: 9, 21:9) and resolution from 1K to 4K.
This model is designed to serve content creators, graphic designers and professional users who need high- honesty images.
The highlight of Nano Banana 2 is the multi-step photo creation process, in which AI will plan, evaluate, adjust and complete the output before displaying. This self- control mechanism helps the final image become more realistic, sharp and accurate, especially in displaying text and small details - an inherent weakness of previous AI photo tools.
This model is integrated with Gemini 3 Pro Image, helping to increase smoothness in the image series with the same character, while improving consistency and cultural awareness.
For example, when users request a spring picnic in Tokyo, AI can accurately reflect the landscape, costumes, and typical lights of the area.
In particular, Google is testing a new Gemini corruption feature, allowing users to change small details in photos such as wallpaper, lighting or costumes... without having to recreate everything.
Along with that, Google also revealed that the upcoming Gemini 3.0 is expected to improve the processing and image understanding performance of Nano Banana 2 to a higher level, paving the way for the new generation of AI images to be able to think, evaluate and create close to humans.
