Google DeepMind introduced Nano Banana Pro, an image creation and editing model also identified as Gemini 3 Pro Image. The system became accessible in Google's Gemini consumer app, its Ads product and Google AI Studio, positioning it as a more capable successor to the earlier image model based on Gemini 2.5 Flash.
The release focuses on generating designed material rather than only standalone pictures. Google says the model can turn source material into prototypes, diagrams and infographics, using Gemini 3 Pro's reasoning and knowledge capabilities to organize information visually. It can also connect to Google Search for grounded inputs such as current weather, sports information or facts used in a recipe illustration.
Text inside images is another central feature. The company says Nano Banana Pro produces more legible wording across multiple languages and can handle material ranging from brief slogans to longer passages. Its examples include posters, typography, calligraphy and packaging translated from English into Korean. The announcement characterizes those results as an improvement, although the supplied evidence does not include comparative benchmark scores or independent testing.
Editing controls are intended to preserve complex source elements while changing a composition. Google demonstrated scenes assembled from several supplied pictures, including people, an animal, clothing and different environmental elements. It says the model can maintain consistency across as many as 14 inputs. Other examples change aspect ratio, lighting or focus while retaining a chosen subject's position.
Those capabilities broaden the intended audience. A designer could build a campaign concept or mock-up, while a teacher or student could request an explanatory diagram. The search connection is meant to support visuals based on recent information rather than only the model's existing knowledge. The announcement nonetheless remains a description from the model's developer and does not establish accuracy for every generated diagram, translation or factual illustration.
Nano Banana Pro follows Nano Banana, which Google released only months earlier for tasks that included photo restoration and miniature-figure generation. The newer version moves from the Gemini 2.5 Flash Image foundation to Gemini 3 Pro, trading on deeper reasoning and a wider set of creative controls.
The launch reflects an effort to combine image synthesis with layout, language and information retrieval in a single workflow. Its practical value will depend on whether real outputs consistently preserve requested identities, render accurate text and represent grounded information correctly. Google provided extensive demonstrations, but the supplied announcement did not detail pricing, usage limits or a formal evaluation methodology.
Google's examples span handwritten notes converted into structured diagrams, plant-care material arranged as an infographic and a recipe represented through sequential illustrations. The announcement uses those demonstrations to connect information organization with creative production, rather than presenting the model solely as a tool for photorealistic scenes.
