Google has announced a major upgrade to its Gemini AI image model, called Gemini 2.5 Flash Image, which enhances photo editing capabilities with greater precision and control based on natural language requests. The new model is designed to preserve facial and object consistency while allowing users to modify images seamlessly, such as changing backgrounds, blending photos, or applying styles. This update has already garnered significant attention on social media, where users have nicknamed the model 'nano-banana' due to its impressive performance and the numerous banana-related hints in its development. Google claims its AI image model is state-of-the-art on benchmarks like LMArena and aims to compete with OpenAI’s image tools and Meta’s licensing of AI models from Midjourney. The company emphasizes safeguards to prevent inappropriate content, including watermarks and metadata identifiers, although concerns about deepfake images persist. The upgrade reflects Google’s effort to close the user gap with rivals like ChatGPT, which currently logs over 700 million weekly users, while Gemini’s monthly users are around 450 million. The company highlights its focus on consumer use cases, such as visualizing home projects, and improving the model’s understanding of complex prompts. The development of this AI tool has sparked curiosity and theories, especially following Sundar Pichai’s cryptic post with three banana emojis, which social media users speculate could be a playful hint or a reference to the new AI capabilities.