Google Launches Gemini 2.0 AI Model, Powering Agents and Enhancing Search

Dec 11, 2024, 03:32 PM

Google has unveiled Gemini 2.0, its latest flagship AI model, introducing significant advancements in AI technology. The model, which includes the experimental version Gemini 2.0 Flash, offers enhanced performance, improved speed, and new multimodal capabilities. Key features include the ability to generate text, images, and speech, as well as support for voice-to-voice interactions and real-time conversation. Gemini 2.0 is designed to power AI agents, with applications in research, web browsing, and gaming. Google has also introduced Project Mariner, an AI agent that can perform tasks within the Chrome browser, and is testing Gemini 2.0 in Google Search for handling complex queries like advanced math and coding. The model was trained on Google's sixth-generation TPU, Trillium, which provides four times the training performance of its predecessor while being more energy-efficient. This release marks a significant step towards more autonomous AI agents, with Google aiming to integrate these capabilities into consumer products. Additionally, Gemini 2.0 Flash is now available in Google AI Studio with 128k tokens, supporting multimodal input and output, and Project Astra capabilities have been detailed, focusing on AI agents for gaming and web browsing.