Nano Banana Pro Launches With 2K–4K Output, Multilingual Text & Improved Accuracy
Google has introduced its latest image-generation system, Nano Banana Pro—also referred to as Gemini 3 Pro Image—arriving just shortly after the debut of its Gemini 3 model. This upgraded version is designed to produce visuals that are more accurate and useful, with Google positioning it as a tool well-suited for educational graphics, diagrams, blueprints, and visual explainers.
One demonstration shows the AI researching how to make cardamom tea, then turning that information into a visual recipe complete with steps and a structured flowchart layout.
Generating readable text inside images has traditionally been difficult for AI models, but Google claims Nano Banana Pro makes significant progress in this area. The company showcases examples of storyboards and illustrated layouts that successfully combine artwork with written instructions, ranging from brief captions to entire paragraphs.
In a companion guide, Google notes that the model may still struggle with tiny text and occasionally misspell words, meaning users might need to refine outputs through additional prompts. The system also supports multilingual text creation, making it possible to quickly produce versions of the same image for different countries or audiences—though Google advises reviewing results carefully for grammar errors or cultural context issues.
Users can now generate visuals in 2K or 4K resolution, and the tool allows targeted editing of specific elements, including lighting, color, focus, camera angle, effects, and more.
