Google brings Nano Banana 2 to Gemini and Search

Google announced the launch of Nano Banana 2, the latest version of its image generation model. Technically named Gemini 3.1 Flash Image, the new model can create more realistic images than its predecessor. It will now serve as the default image generation model across the Gemini app, specifically powering its Fast, Thinking, and Pro modes. This update follows the original release of Nano Banana in August 2025—which prompted users to generate millions of images, particularly in countries like India—and the subsequent release of Nano Banana Pro in November 2025. While the Pro model was designed for highly detailed, specialized tasks, Nano Banana 2 aims to deliver high-fidelity images at faster speeds.

The technical specifications of the new model include:

• Resolution: Ranging from 512px to 4K, across various aspect ratios.
• Character consistency: Capable of maintaining character consistency for up to five characters in a single workflow.
• Object fidelity: Capable of maintaining fidelity for up to 14 objects in one workflow.

Google is integrating the model across its broader ecosystem. Beyond the Gemini app, Nano Banana 2 is now the default model for Google’s video editing tool, Flow. In Search, the model will become the default for results generated via Google Lens and in AI Mode across 141 countries, accessible on both desktop and mobile. For subscribers on Google’s higher-end plans, Google AI Pro and Ultra, the option remains to use Nano Banana Pro for specialized tasks by regenerating images. For developers, Google is making Nano Banana 2 available in preview through the Gemini API, Gemini CLI, and Vertex API, as well as through AI Studio and its development tool, Antigravity, which was released in November 2025.

The launch also marks a push toward standardizing safety and content provenance. All images generated with Nano Banana 2 will feature a SynthID watermark, Google’s watermarking technology for AI-generated images. Google reported that since launching SynthID verification in the Gemini app in November 2025, users have verified images over 20 million times. Additionally, the generated images are interoperable with C2PA Content Credentials, an industry standard for content provenance supported by a coalition of technology companies including Adobe, Microsoft, OpenAI, and Meta.

Why it matters

Google is standardizing its image generation capabilities across consumer apps and developer tools, while embedding safety features like SynthID and C2PA to address content provenance.