Google Search AI Mode gets a substantial upgrade with Gemini 3

We already know that Gemini 3 is a new, powerhouse model, but the biggest question with these launches is always: “When will I actually see it in the products I use?” Usually, the answer for Google Search is “months from now.” Not this time.

In a blog post detailing the integration, Google announced that Gemini 3 is coming to Google Search starting today. This marks the first time Google has launched a new frontier model into Search on day one. But it’s not just about better text answers. The most mind-blowing update here is something Google calls “Generative UI.”

Search can now build apps for you on the fly

We talked in our other post about the fact that Gemini 3 is good at Vibe Coding, but it seems Google is applying that directly to the Search results page as well. With Generative UI, Gemini 3 in Google Search AI Mode doesn’t just retrieve information; it can dynamically code and render interactive tools, tables, and simulations specifically for your query.

“When the model detects that an interactive tool will help you better understand the topic, it uses its generative capabilities to code a custom simulation or tool in real-time and adds it into your response.”

Google gave two wild examples of how this works:

1. The Three-Body Problem: If you ask about the physics behind this concept, instead of just text, Search will generate an interactive simulation where you can manipulate variables and watch the gravitational interactions happen right in your browser.
2. Mortgage Loans: If you are researching loans, it can code a custom-built interactive calculator that lets you compare specific options based on the data you asked about.

It’s essentially building a mini-app for you, in seconds, just to answer your question.

“Query Fan-Out” and Smarter Routing

Under the hood, Gemini 3 is also changing how Google searches. Google is upgrading its “query fan-out” technique. Because Gemini 3 understands nuance and intent so much better, it can perform a wider range of searches to find content that previous models might have missed. Essentially, it’s better at connecting the dots on complex questions.

How to use it right now

Here is the catch (there’s always one): This is currently rolling out to Google AI Pro and Ultra subscribers in the U.S. If you have a subscription:

1. Go to Google Search.
2. Enter AI Mode.
3. Select “Thinking” from the model drop-down menu.

Google says this will eventually roll out to everyone in the U.S. (with usage limits), and in the coming weeks, they will introduce automatic model selection. This means Search will be smart enough to route simple queries to fast models and hard questions to Gemini 3 automatically, without you needing to toggle anything.