• Google’s Nano Banana 2 AI image generator is popping up in preview
• The new model promises smarter, self-correcting image generation
• The new model mimics a multi-step human workflow

Google is gearing up for another bite at advanced AI image generation with Nano Banana 2. The follow-up to the original Nano Banana model will be part of the Gemini app, according to the preview that unexpectedly appeared and was shared on X.

Nano Banana 2 looks to bring better angle and viewpoint control of images, with more precision in coloring. There will also be the option to fix text in the image without messing with the rest of the output.

So, Nano Banana 2 will be more than just a resolution bump. Nano Banana 2’s biggest shift may be how it thinks. The leaked previews suggest the model now adopts a multi-step workflow where it will plan the image before making it, and analyze it for mistakes, fixing them and repeating the process until it’s ready.

That kind of built-in self-correction is new for Google’s image tools. Behind the quirky fruit nickname, Google appears to be treating AI image generation like a real design assistant, one that sketches rough drafts, spots flaws, and only hands you the final product after making sure it isn’t terrible.

The images shared by those with access to the preview show cleaner lines, sharper angles, and fewer of the telltale errors of AI images.

“Nano Banana Pro” has started popping up in GitHub commits and code references, suggesting that Google is already preparing a more premium-tier version of the model for high-end or high-res tasks.

Appealing AI images

In the wild, Nano Banana 2, internally known as GEMPIX 2, has started appearing in places beyond the main Gemini app as well. Testers have found traces of the model in experimental tools like Whisk Labs, part of Google’s ongoing push to weave AI creativity into all of its tools. If the past is any indication, this multi-surface rollout will follow the same playbook as the original Banana drop, with users suddenly discovering their photos look better.

The fact that Nano Banana 2 goes through a self-correction loop before delivering the final image marks a philosophical shift. It’s not just reactive anymore. Google is teaching its model to notice its own mistakes and fix them proactively. That sounds small, but it mirrors the human creative process more than past AI tools have.

Better scene understanding, angle control, and text clarity should mean you’re more likely to get an image you actually want. Considering how Nano Banana went viral for making action figure versions of people that looked like real products, it will be worth watching to see how well Nano Banana 2 carries out similar images set in the real world. Judging by some of the shared images and their consistency in holding on to that character likeness, you’ll see some really realistic, unreal depictions of people and places very soon.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

Eric Hal Schwartz is a freelance writer for TechRadar with more than 15 years of experience covering the intersection of the world and technology. For the last five years, he served as head writer for Voicebot.ai and was on the leading edge of reporting on generative AI and large language models. He’s since become an expert on the products of generative AI models, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google Gemini, and every other synthetic media tool. His experience runs the gamut of media, including print, digital, broadcast, and live events. Now, he’s continuing to tell the stories people want and need to hear about the rapidly evolving AI space and its impact on their lives. Eric is based in New York City.