KUALA LUMPUR – Google has officially introduced Nano Banana 2, also known as Gemini 3.1 Flash Image, its latest artificial intelligence (AI) model designed to significantly improve consistency and precision in visual generation.
The new model marks a major leap forward from its predecessor, specifically addressing common AI hurdles such as blurred text and spelling errors within generated images. By leveraging Gemini’s real-world knowledge base and real-time data from web searches, Nano Banana 2 allows users to create highly accurate infographics, diagrams, and educational charts.
A standout feature of this update is its enhanced visual consistency. Nano Banana 2 is capable of maintaining up to 14 distinct objects across a series of generations, ensuring that characters or primary subjects remain stable and recognizable throughout multiple images.
Furthermore, the model now supports image generation in resolutions up to 4K and offers various aspect ratios to suit diverse project requirements. Google has also refined the model’s ability to handle text translation and localization directly within visuals, providing a level of accuracy previously unseen in generative AI.
“Nano Banana 2 is built to generate more consistent images and more precise text,” according to reports. This makes it an invaluable tool for creators who require contextual visuals, such as data-driven content and complex graphic designs.
The model is currently being rolled out in phases across the Gemini app and other Google services. It is set to replace previous iterations as the default choice for image generation, signaling Google’s commitment to leading the next wave of functional and professional AI-driven creativity.
