Google launches Nano Banana Pro with sharper visuals and better text rendering

Nano Banana

Google has rolled out Nano Banana Pro, its upgraded AI image model built on Gemini 3 Pro, and it is a noticeable step up from the hyperrealistic figurine style that went viral back in September. The company says the new version can turn simple prompts into studio-quality designs with far more control, better accuracy and world knowledge, and text that actually looks like text.

Nano Banana Pro is available globally from today and anyone can try it for free inside the Gemini app by choosing Create image with the Thinking model. Google says the model is now smart enough to produce context-rich infographics, diagrams and visualisations that can show dynamic information like weather or sports. Its ability to render clean, legible text directly within the image opens the door to posters, invites or multi language layouts that do not fall apart on inspection.

The model also lets you blend up to 14 images and as many as five people into a single scene. Editing has been pushed even further. You can now select a specific region and make local changes, adjust camera angles, switch lighting from day to night, tweak colours, add bokeh or refocus the entire frame. Output goes up to 4K across a wide range of aspect ratios.

Importantly, every image created or edited using Nano Banana Pro will carry C2PA metadata. It will not fix the deepfake problem overnight, but it is a necessary step towards clearer labelling as more platforms commit to transparency. TikTok, for instance, has also begun using C2PA to mark AI generated visuals.

Free users get a limited quota, while Google AI Plus, Pro and Ultra subscribers get expanded access. The model also works inside AI Mode in Search for Pro or Ultra users in the US, and globally through NotebookLM.