- Google’s Nano Banana 2 AI picture generator is popping up in preview
- The brand new mannequin guarantees smarter, self-correcting picture technology
- The brand new mannequin mimics a multi-step human workflow
Google is gearing up for an additional chunk at superior AI picture technology with Nano Banana 2. The follow-up to the unique Nano Banana mannequin will probably be a part of the Gemini app, in response to the preview that unexpectedly appeared and was shared on X.
Nano Banana 2 appears to convey higher angle and viewpoint management of photos, with extra precision in coloring. There may also be the choice to repair textual content within the picture with out messing with the remainder of the output.
So, Nano Banana 2 will probably be greater than only a decision bump. Nano Banana 2’s largest shift could also be the way it thinks. The leaked previews counsel the mannequin now adopts a multi-step workflow the place it’ll plan the picture earlier than making it, and analyze it for errors, fixing them and repeating the method till it is prepared.
That type of built-in self-correction is new for Google’s picture instruments. Behind the quirky fruit nickname, Google seems to be treating AI picture technology like an actual design assistant, one which sketches tough drafts, spots flaws, and solely palms you the ultimate product after ensuring it is not horrible.
Nano Banana 2 is now in Preview 👀presently on Media IO pic.twitter.com/VNmQM3PAFPNovember 8, 2025
The photographs shared by these with entry to the preview present cleaner traces, sharper angles, and fewer of the telltale errors of AI photos.
“Nano Banana Professional” has began popping up in GitHub commits and code references, suggesting that Google is already making ready a extra premium-tier model of the mannequin for high-end or high-res duties.
Interesting AI photos
Within the wild, Nano Banana 2, internally often known as GEMPIX 2, has began showing in locations past the principle Gemini app as properly. Testers have discovered traces of the mannequin in experimental instruments like Whisk Labs, a part of Google’s ongoing push to weave AI creativity into all of its instruments. If the previous is any indication, this multi-surface rollout will observe the identical playbook as the unique Banana drop, with customers instantly discovering their images look higher.
The truth that Nano Banana 2 goes via a self-correction loop earlier than delivering the ultimate picture marks a philosophical shift. It’s not simply reactive anymore. Google is instructing its mannequin to note its personal errors and repair them proactively. That sounds small, but it surely mirrors the human inventive course of greater than previous AI instruments have.
Higher scene understanding, angle management, and textual content readability ought to imply you’re extra prone to get a picture you really need. Contemplating how Nano Banana went viral for making motion determine variations of those who regarded like actual merchandise, it will likely be price watching to see how properly Nano Banana 2 carries out comparable photos set in the true world. Judging by among the shared photos and their consistency in holding on to that character likeness, you may see some actually lifelike, unreal depictions of individuals and locations very quickly.
Comply with TechRadar on Google Information and add us as a most popular supply to get our knowledgeable information, critiques, and opinion in your feeds. Make certain to click on the Comply with button!
And naturally you may also observe TechRadar on TikTok for information, critiques, unboxings in video kind, and get common updates from us on WhatsApp too.
The most effective enterprise laptops for all budgets
