Google Launches Gemini 2.0: A New Era for AI Agents
Google has unveiled Gemini 2.0, its latest AI model designed for a more agentic approach to information processing. CEO Sundar Pichai emphasizes the shift from organizing and understanding information (Gemini 1.0) to making it more useful (Gemini 2.0).
The first available model, Gemini 2.0 Flash, surpasses Gemini 1.5 Pro in key benchmarks like code, factuality, math, and reasoning, all while operating at twice the speed. It boasts multimodal output, including images and multilingual audio, and can utilize tools like Google Search and code execution. Developers can access the experimental version via AI Studio and Vertex AI, with general availability slated for January.
For users of the Gemini app, the new model promises a more helpful Gemini assistant. Both Gemini and Gemini Advanced users can experiment with a chat-optimized version on gemini.google.com. Mobile app access is coming soon. Gemini 2.0 is also being tested in Search’s AI Overviews to handle complex queries, including math equations and coding. Wider availability is expected early next year.
This new model will be integrated into other Google products in the near future. For more on Google's AI advancements, see Gemini 2.0: Google's New AI Agents for Everyday Tasks and Gemini Advanced Debuts "Deep Research" Feature. You can also explore related updates on Gemini UI Refresh: Web and Android Updates.