Google just made the best version of their image-creating AI (Nano Banana) waaay
By Christophermrawlings · March 21, 2026 · Curated by George's Blog
Google just made the best version of their image-creating AI (Nano Banana) waaay harder to use. Kinda insane. The good part is if you know what you’re doing you can still make amazing product images with it.
We tested the new Nano Banana 2 against Nano Banana Pro for product images.
Spoiler - Nano Banana 2 came out SO much worse.
Its not useless. Its just worse in the important places where it counts for Amazon images.
The issue begins with the fact the Google Gemini app now makes users start with Nano Banana 2 even if they’re paying for the good plan.
This means instead of getting the best images right there, users are starting with a weaker version of the AI and have to try and fix their images afterward.
That matters because if the first image gets the product wrong (like the wrong color or shape or some weird design detail) then trying to fix it up later usually isn’t enough to save things.
It’s like building a house on bad foundations.
The good news is there’s still a way around it - instead of using the Gemini app go into Google AI Studio and make an API key. Then open the Playground and switch the model to Nano Banana Pro.
Its a bit more annoying but it gets you seriously better results.
Comment "Video" and i will send over the full video of me breaking this down