Krux
February 27, 2026
Google's Image AI Now Searches the Web Before Drawing
Published: February 27, 2026 at 12:35 AM
Updated: February 27, 2026 at 12:35 AM
100-word summary
Google just released Gemini 3.1 Flash Image (codenamed Nano Banana 2), and the standout feature isn't sharper pixels. It can search the web before generating an image, grounding visuals in real locations, events, or objects. Need a poster with accurate Tokyo street signs or a product shot mimicking a viral aesthetic? The model pulls reference imagery first, then renders. It also handles extreme aspect ratios (8:1 banners, anyone?) and finally fixes AI's nemesis: readable text in multiple languages. The catch? Outputs still vary, and you'll need a paid API key to try it in Google AI Studio.
What happened
Google just released Gemini 3.1 Flash Image (codenamed Nano Banana 2), and the standout feature isn't sharper pixels. It can search the web before generating an image, grounding visuals in real locations, events, or objects. Need a poster with accurate Tokyo street signs or a product shot mimicking a viral aesthetic? The model pulls reference imagery first, then renders. It also handles extreme aspect ratios (8:1 banners, anyone?) and finally fixes AI's nemesis: readable text in multiple languages. The catch?
Why it matters
Outputs still vary, and you'll need a paid API key to try it in Google AI Studio.