Google on 26 August began rolling out Gemini 2.5 Flash Image, an image-generation and editing model internally codenamed “Nano Banana.” The system is now live in the Gemini mobile app, Google AI Studio and the Gemini API, giving users the ability to create, refine and remix visuals directly within those services. Executives from Google DeepMind, including Chief Executive Officer Demis Hassabis, said the model improves character consistency, multi-turn editing and stylistic transfers while matching or surpassing photorealism benchmarks. According to the community-run Image Edit Arena, the release has already attracted roughly five million votes, with more than half cast for the new model, placing it at the top of the leaderboard. Google says outputs are automatically labeled with an AI watermark and the company’s invisible SynthID signature to help identify synthetic content. The upgrade sharpens competition with established image generators as Google pushes to embed advanced multimodal capabilities across its consumer and developer platforms.