Google has announced Gemini 2.0, its most capable AI model yet, which is designed to be more “agentic,” meaning it can understand the world, think ahead, and take action. Gemini 2.0 will be integrated into Google products like Search, enabling more complex and multi-step questions, including advanced math, multimodal queries, and coding. The new model builds on the success of Gemini 1.5 and features enhanced performance, multimodal output like native image generation, and the ability to use tools like Google Search and code execution. Google is also exploring the practical applications of AI agents with prototypes like Project Astra, a universal AI assistant, and Project Mariner, which explores human-agent interaction in a web browser. These advancements are underpinned by Google’s custom AI hardware, Trillium, which is now generally available to ...