Google Launches Gemini 3, Promising Advanced AI Capabilities

Google has officially launched its latest artificial intelligence model, Gemini 3, which aims to revolutionize user interaction by providing enhanced capabilities across various applications. This release marks a significant milestone as it is being made available to Search users from day one, expanding access to a broader audience.

The key feature of Gemini 3 is its ability to understand context and intent more effectively than previous iterations. Google asserts that this model can “learn anything” and “build anything,” enabling users to engage with AI in more meaningful ways. For instance, users can now translate handwritten recipes into a digital family cookbook, making it easier to share culinary traditions.

In the realm of education, Gemini 3 facilitates learning by processing academic papers and long video lectures. It can generate interactive tools such as flashcards and visualizations, tailored to help users grasp complex topics. Furthermore, it provides personalized analysis for activities like sports; for example, analyzing pickleball matches to enhance performance through targeted training plans.

New Features and Developer Tools

The integration of Gemini 3 into the AI Mode of Google Search introduces innovative user experiences. This includes features such as “immersive visual layouts” and “interactive tools and simulations,” which are designed to respond directly to user queries with greater sophistication.

For developers, Gemini 3 delivers impressive performance metrics. It has achieved a score of 1487 Elo on the WebDev Arena leaderboard and a 54.2% on Terminal-Bench 2.0, surpassing its predecessor, Gemini 2.5 Pro, in various benchmarks. With these enhancements, developers are expected to find Gemini 3 more beneficial for creating websites, games, and applications. Access to these tools is now available within AI Studio.

An experimental feature, known as Gemini Agent, also debuts with this launch. This tool is designed to perform multi-step tasks by connecting with Google applications, streamlining tasks such as managing calendars and emails. Users can instruct Gemini to automate travel arrangements, for example, by asking it to research and book a rental vehicle based on email details. Initially, Gemini Agent will be available to subscribers of Google AI Ultra starting today.

Gemini 3 can be accessed through the Gemini app, with availability extending to Google AI Pro and Ultra subscribers in AI Mode within Search. Developers can utilize the Gemini API in AI Studio from this launch date as well.

With the introduction of Gemini 3, Google continues to push the boundaries of artificial intelligence, providing tools and capabilities that promise to enhance user experience and foster greater interaction with technology.