Google announced on the 11th the launch of its latest AI model, Gemini 2.0 ("2.0"), marking a year since the introduction of Gemini 1.0 in December last year. The 1.5 version was also released in February this year.

According to Google, 2.0 is the most advanced AI model it has introduced to date. It is a multimodal model equipped to handle text, images, and video and is optimized for the era of AI agents. Key features include faster responses, natural conversations, and enhanced multimodal capabilities, allowing it to serve as an effective AI agent for users.

The model is built on Google's proprietary 6th-generation TPU chip, Trillium, enabling not only the structuring and understanding of information but also making information more useful. Starting today, 2.0 is available to developers and testers, with plans to integrate it rapidly across all Google products, beginning with Google Search.

Google has incorporated 2.0 into its "Project Astra," launched in May, to improve natural conversational abilities, response speed, and memory retention. Project Astra is Google’s vision for a future AI assistant capable of seeing, listening, and conversing like a human while acting as a personal assistant. Demis Hassabis, CEO of Google DeepMind, explained, "Gemini 2.0 offers a groundbreaking agent-based experience with enhanced interaction, rapid responses, and the ability to handle complex tasks."

One of the models in the 2.0 lineup, 2.0 Flash, is now available for experimentation through Google AI Studio for developers and Vertex AI for enterprise users. The Flash model is a streamlined version of the Pro model, part of the Gemini lineup categorized by parameter sizes, including Ultra, Pro, and Nano. The Flash series has been available since the 1.5 version.