Google launched “Nano Banana 2,” a latest iteration of its artificial intelligence image generation and editing tool, on Tuesday, offering users access to “pro-level” quality images at a significantly faster speed, according to a press release from the company.
The new tool, officially named Gemini 3.1 Flash Image, builds upon the high-quality inference capabilities of “Nano Banana Pro,” released in November 2025. Google stated that Nano Banana 2 combines this quality with accelerated generation speeds, making premium features previously available only to paid subscribers now accessible to a wider audience.
Nano Banana 2 supports the translation of text within images into different languages and localizes content, leveraging real-time web searches and Google’s Gemini’s knowledge base to accurately depict visuals based on current information. This functionality is expected to be particularly useful in creating infographics and data visualizations, according to Google.
The tool maintains consistency for up to five characters and 14 objects within a single image, preventing distortions in character appearance – a feature designed to aid in the creation of illustrated books and comics. It supports resolutions ranging from 512 pixels for rapid iteration to 4K ultra-high definition, and accommodates various aspect ratios, including 4:1 and 1:8.
Google has as well incorporated a feature allowing professionals to adjust the model’s “reasoning level,” enabling more complex tasks to benefit from thorough inference processes. The update integrates Nano Banana 2 as the default model within the Gemini app and other Google products. It will also serve as the default image generator in Google Search across 141 countries, accessible through Google Lens and AI modes. Users with Google AI Pro and Ultra subscriptions will still have the option to utilize Nano Banana Pro.
Developers will gain expanded API support through Gemini API, Gemini CLI, Vertex API, AI Studio, and Antigravity, a development tool released in November 2025. All generated images will include Google’s AI-generated image watermark, “SynthID,” and are compatible with the C2PA content credentials framework, a coalition including Adobe, Microsoft, OpenAI, and Meta. Google reported that SynthID verification has been utilized over 20 million times since its introduction in November.
