Google has officially launched the Nano Banana 2 image model. The new model is built on Gemini 3.1 Flash Image and boasts faster response times and stronger understanding capabilities compared to its predecessor, with more accurate parsing of user prompts.
According to Google, Nano Banana 2 has significantly improved in terms of generation consistency, while also fixing common issues in previous versions such as garbled Chinese characters, semantic confusion, or strange artifacts in the image.
In terms of performance, the Nano Banana 2’s image output resolution has been increased from 2K to a maximum of 4K, enabling it to handle more complex prompts, support keeping up to 5 character images consistent across multiple images, and blend up to 14 elements into a single image.
Subsequently, Nano Banana 2 will gradually be rolled out to Google Gemini, Google Search, Google AI Studio, Google Flow (Note: Google’s AI creative tool), paid API services, and Google Ads, replacing the old version of Nano Banana.
