Google released an updated version of Gemini 2.0 artificial intelligence, which will become the basis for all the company’s products. The release took place 10 months after the launch of previous version 1.5.
Demis Hasabis, CEO of Google DeepMind and leader of all the company’s artificial intelligence projects, emphasizes the versatility of the new model. According to him, Gemini 2.0 Flash demonstrates efficiency at the level of the current Pro version at the same speed and cost of computing.
The updated version has the ability to create audio and images, and has expanded multimodal features. These improvements lay the groundwork for the development of agent-based artificial intelligence — bots that can perform tasks on behalf of users.
«We see 2025 as the true beginning of the agent era, and Gemini 2.0 is the foundation for that,» Hasabis said.
The company presented a number of experimental projects. Project Astra is a visual system that helps to identify objects, navigate in space, and find lost items. Project Mariner is a Chrome extension that can control the browser. Jules specializes in finding and fixing code errors. A separate agent based on Gemini 2.0 helps users improve their gaming skills.
Hasabis recognizes the need to address both traditional problems — computing performance, efficiency, and cost, as well as new challenges related to the security of autonomous agents.
«We need new security solutions, such as testing in protected environments. This will be important for testing agents, instead of real-world testing… it will be more useful, but the risks will increase,» he explains.
Currently, only the basic version of the — Gemini 2.0 Flash model is available, which can be tested in the Gemini web application. Early next year, the update will be available on the company’s other platforms.
Source: The Verge
Spelling error report
The following text will be sent to our editors: