Google has just put itself on the table with the arrival of Gemini 2.0, the new version of its artificial intelligence model whose main objective is to allow the creation and use of what we call “AI Agents“, and this concept is something we will be talking about a lot in the immediate future, says the company.
AI agents, in essence, are assistants who can do things without us having to intervenejust give him orders. For example, Google introduced an experimental Chrome extension that can use the browser on our behalf, as well as another bot that can more quickly detect errors and problems in the code and help developers fix them. This is just the beginning.
This is Gemini 2.0, Google's new model for “AI agents”
Specifically, Google today launched the Gemini 2.0 Flash model, which is the smallest and least powerful model of all. Nevertheless. Gemini 2.0 Flash is basically as good as the current Pro modelaccording to Demis Hassabis, CEO of Google DeepMind.
One of the main novelties of this model is that it is more versatile and more efficient than previous models; For example, Gemini 2.0 Flash can generate multilingual images and audio natively. According to Hassabis:
“[Gemini] 2.0 Flash even outperforms 1.5 Pro in key benchmark tests, with twice the speed and also comes with new features. [Además] “It's natively integrated with tools like Google search or code execution, as well as third-party user-defined functions.”
But the biggest new thing in Gemini 2.0 is the artificial intelligence agents. In addition to the products and examples mentioned above, Google has also demonstrated the usefulness of this concept with an agent for video games that helps the user play better. For the company, this type of product and concept is one of the most important we will see in terms of artificial intelligence from 2025 onwards.
Google has also demonstrated the capabilities of multimodal reasoning in the past, as recalled in the official announcement of Gemini 2.0. This year, the company unveiled Project Astra, the evolution of virtual assistants which can analyze our environment to, for example, find lost objects, which we have already seen during a demonstration of this technology. Astra also received improvements thanks to Gemini 2.0, including better dialog with support for multiple languages, better memory (meaning it can remember more things), and the ability to use more of tools, including Lens, Maps and Google Search.
Gemini 2.0 Flash Experimental is now available for early adopters. This is an experimental version of this new AI model, but Google nevertheless assures that it is its most advanced and high-performance model to date. Developers can also start using this new model via the Gemini API.
Cover image | Google (with editing)
More information
In Xataka Android | Gemini Live is Google's AI that finally converses and speaks Spanish. These are the five most useful things it can help you with
In Xataka Android | Gemini is the ultimate detector for detecting fraudulent and spam-heavy SMS messages. You don't even need to switch apps
