Mysterious ‘Nano-Banana’ Undertaking Revealed to Be Google’s Newest Picture Editor

Google simply upgraded its AI picture mannequin, and it really seems to be a reasonably important step up. The corporate rolled out Gemini 2.5 Flash Picture at the moment, a significant refresh that guarantees smarter and extra versatile picture era.

The upgraded mannequin permits customers to subject pure language prompts to not solely generate pictures but additionally merge present photographs and make extra exact edits with out creating bizarre distortions. It additionally faucets into Gemini’s “world data” to raised perceive what it’s producing. 

This improve comes as Google tries to shut the hole with the trade chief, OpenAI. Up to now, picture era has been a significant driver for AI. ChatGPT utilization skyrocketed in March when the corporate launched its GPT-4o native picture generator. The viral Studio Ghibli memes generated by the mannequin resulted within the firm’s GPU models melting, based on OpenAI CEO Sam Altman. ChatGPT at the moment has over 700 million weekly users. By comparability, Google CEO Sundar Pichai revealed on the corporate’s July earnings name that Gemini had 450 million monthly users, nonetheless trailing behind ChatGPT. 

With its newest replace, Google says it’s solved certainly one of AI’s largest complications. Till now, retaining characters or objects constant throughout a number of edits has been a significant problem for AI picture mills.

“Now you can place the identical character into completely different environments, showcase a single product from a number of angles in new settings, or generate constant model property, all whereas preserving the topic,” the corporate wrote in a blog post.

Google says customers can now make very particular tweaks with only a immediate. For instance, customers can blur the background of a picture, take away a stain from a T-shirt, change a topic’s pose, and even add coloration to a black-and-white picture.

Even earlier than its official launch, the brand new mannequin was turning heads on the crowdsourced analysis platform LMArena, the place it appeared anonymously underneath the identify “nano-banana.” One X user shared how they used nano-banana to alter Altman’s shirt in a photograph. The consequence was surprisingly good. Right this moment, Google stepped ahead and claimed possession of the mannequin, revealing that nano-banana was the truth is Gemini 2.5 Flash Picture.

Along with being obtainable on the Gemini app, the brand new mannequin is now accessible to builders by the Gemini API, Google AI Studio, and Vertex AI. Google has already constructed a number of template apps that make use of the brand new mannequin on Google AI Studio, the corporate’s coding AI assistant, and mentioned customers can vibe code on prime of them.

The corporate additionally mentioned some builders have already experimented with the app to see how it might be helpful in real-world eventualities, like creating actual property itemizing playing cards, worker uniform badges, and product mockups.