Google has created a new model for image generation — Nano Banana

Google has introduced Nano Banana — a preview version of the gemini-2.5-flash-image-preview model, part of the Gemini 2.5 ecosystem. Despite its playful name, this is a serious tool that could become the new standard in visual AI.

Nano Banana supports multimodal queries: you can input text, images, or a combination of both, and receive an image, JSON, or structured response as output.

Key features include:

— image generation from text descriptions
— local editing of individual objects or scenes
— preservation of character visual consistency
— style transfer
— up to 30x faster performance than previous Google models. The model is already available in the free Gemini App, via the paid Gemini API on Google Cloud, in Vertex AI for enterprise clients, and through the Krea editor with a subscription.

Nano Banana is targeted at digital agencies, game studios, e-commerce, marketing, and visual storytelling. Thanks to its stability in faces and scenes, Nano Banana is well-suited for creating characters, AI influencers, and highly detailed content.

While Nano Banana is currently available in preview mode, it is already ready for commercial use.

Source: nanobanana