Google has presented an updated version of its flagship artificial intelligence model – Gemini 2.0 Flash. It works twice as fast as the previous version and will become the basis for AI agents that will help users.

The new model is capable of creating images and audio in different languages, facilitating Google searches, helping with code writing, and even supporting users during gaming sessions.

Google launches Project Mariner based on Gemini 2.0, an experimental Chrome extension that will allow an AI agent to control the browser independently. New agents are also introduced: Jules – for finding and fixing errors in the code, and an agent based on Gemini 2.0 that can monitor the user’s screen and help them during video games.

Gemini 2.0 will become the basis for almost all Google products. It will be able to more accurately summarize search results in the feed and take into account more nuances, the developers assure.

The new model will be integrated into the chatbot and will later become the basis for AI functions in Workspace and other company services. Google plans to combine the maximum number of features in one model instead of creating many separate solutions.

Gemini 2.0 is already available in the web version of the chatbot. In early 2025, the model will be rolled out to other Gemini platforms and all Google products.