Announcements or awards from leading AI conferences such as NeurIPS, ICML, or CVPR

Google Launches Gemini 2.0 Flash AI, Powers Astra and Mariner Projects

Dec 11, 2024, 07:17 PM

Google has unveiled Gemini 2.0, its most advanced artificial intelligence model to date, marking a significant leap in the company's AI capabilities. Developed by Google DeepMind, Gemini 2.0 is a multimodal AI model capable of handling text, images, audio, and video inputs and outputs. A key feature is the 'Gemini 2.0 Flash', which offers doubled speed and enhanced real-time capabilities, including 3D spatial understanding. Gemini 2.0 also boasts a two-million token context window for improved reasoning. The model powers new tools such as Project Astra, an AI assistant integrating various Google tools into a single agent, and Project Mariner, which allows AI to navigate the web like a human, interacting with websites by clicking, scrolling, and filling out forms. The release includes the Multimodal Live API, enabling developers to create conversational AI applications with voice and video capabilities. Gemini 2.0 introduces features like Deep Research, an AI research assistant that can perform online research under user supervision, and demonstrates advanced coding abilities. Integrating with Google services like Search and Maps, Gemini 2.0 Flash supports native audio and image generation. The announcement has been well-received, with industry experts noting its potential to redefine generative AI and usher in a new era of AI agents. Google’s stock soared to a record high following the CEO's announcement of the new AI tool. The company emphasizes that Gemini 2.0 prioritizes safe and responsible AI use and plans to expand testing of its capabilities to more users soon.