Google Unveils Gemini 2.0: Faster, Multimodal AI Model Powers New AI Agents

Dec 11, 2024, 04:54 PM

Google has unveiled Gemini 2.0, its most advanced AI model to date, introducing significant upgrades in speed and performance. According to Jeff Dean, Google's Senior Fellow, "The Gemini 2.0 Flash model is almost universally better than the Gemini 1.5 Pro model but with the latency and speed of the Gemini 1.5 Flash model." The new model features multimodal capabilities, including native image and audio generation, and is twice as fast as its predecessor. Gemini 2.0 is designed to power a new generation of AI agents, including Project Astra, a universal AI assistant; Project Mariner, capable of taking actions in Chrome; and Jules, a code agent. Sundar Pichai, Google's CEO, stated that Gemini 2.0 marks a "new agentic era" in AI development, bringing the industry closer to the vision of a universal assistant. Developers can now access Gemini 2.0 Flash via the Multimodal Live API in Google AI Studio and Vertex AI, with general availability expected early next year. The model supports real-time multimodal API, enabling text-to-speech, image generation, and advanced video understanding. Built on Trillium hardware, Gemini 2.0 has already been integrated into enhanced Google Search and Maps. The model debuts as the third overall on the LM Arena leaderboard, with top rankings in various subcategories. While the model is impressive, it is currently in an experimental phase, with some features like image and audio generation available only in private preview.