'Nano-Banana' AI: Google DeepMind brings world’s 'top-rated' image editor to Gemini app

Google is giving Gemini users a big creative upgrade. The company has announced that its new image editing model, developed by DeepMind, is now integrated directly into the Gemini app — making it easier to tweak photos while keeping people and pets looking like themselves.

In a blog post, Google says this model is already the highest-rated image editor in the world, thanks in part to its ability to preserve likeness across edits. That’s been a sticking point for AI art tools, where a small distortion in a familiar face or pet can feel jarring. Now, whether you’re giving yourself a 1960s beehive hairstyle or dressing your chihuahua in a tutu, Gemini is designed to keep the details consistent.

The process is simple: upload a photo, describe what you want changed, and Gemini generates the edit while holding on to the original look. You can even combine multiple photos — say, putting yourself and your dog on a basketball court — or change the setting of a room to preview new wallpaper. Once finished, the image can be fed back into Gemini to generate a video from your creation.

Google is also highlighting advanced use cases. Multi-turn editing lets users gradually build out a scene, like painting a room, then adding furniture and decorations in steps. Style transfer makes it possible to apply textures or patterns from one photo onto objects in another, such as designing boots inspired by flower petals.