Google drops Nano Banana 2: an AI that writes text on images and never loses track of characters. Is anyone an artist now?

Google has rolled out an update for its image generation model, Nano Banana 2. According to the developers, it takes the best of the previous version but kicks things up a notch by doing it all at lightning speed. The update now taps directly into Google Search. So, if you want a picture of a real-world object or landmark, the AI doesn't just guess based on its dataset—it hits up Google for the latest, most accurate references.

One of the biggest headaches for every generative AI out there has been rendering text and symbols, which usually end up as a mess of unreadable scribbles. Nano Banana 2 claims to have cracked the code. The model can now weave meaningful text into an image and even translate it into other languages while keeping the original font and style perfectly intact.

On top of that, the model now has a memory for character and object details. Engineers at Google DeepMind say it can keep track of up to 5 different characters and maintain the consistency of 14 individual objects within a single scene. That means you can finally create full-on storyboards and comics where your main character doesn't randomly change faces from one panel to the next.

The model is already becoming the new standard in the Gemini app (taking over for the Pro version on quick tasks), getting integrated into Search (within AI Mode and Lens), and will be used in Google Ads to whip up campaign copy on the fly. Developers can also tap into the engine via API for their own projects, with support for resolutions all the way up to 4K.

On the safety front, Google is sticking with its SynthID watermarking, which bakes an invisible digital signature into every AI-generated image. The company reports that the AI content verification feature in Gemini has already been used over 20 million times.

If AI gets to the point where it can flawlessly render any object, pull real-time context from search, and translate text on the fly—does that finally pull the rug out from under illustrators and copywriters for good? Drop your thoughts in the comments.