We are excited to announce significant enhancements to the performance and availability of our most popular Gemini features, as well as the introduction of a new feature that will make Gemini even more personalized and useful.

2.0 Flash Thinking Experimental: Enhanced Features and Expanded Context Window

Today, we are rolling out an upgraded version of our 2.0 Flash Thinking Experimental model, which now supports additional features such as file upload. This model, designed to break down prompts into a series of steps to enhance its reasoning capabilities and deliver better responses, has been optimized for improved efficiency and speed. Furthermore, Gemini Advanced users will now have access to a 1M token context window with 2.0 Flash Thinking Experimental, enabling them to tackle more complex problems than ever before by exploring and analyzing large amounts of information.

Deep Research: Smarter and More Accessible

In December, we introduced a new product category in Gemini with Deep Research, a personal AI research assistant that saves users hours of time by searching and synthesizing information from across the web in just minutes. Deep Research also helps users discover sources from across the web that they may not have otherwise found.

Today, we are upgrading Deep Research with the integration of Gemini 2.0 Flash Thinking Experimental. This enhancement expands Gemini’s capabilities across all research stages, from planning and searching to reasoning, analyzing, and reporting, resulting in higher-quality, multi-page reports that are more detailed and insightful. Additionally, Gemini now provides a real-time glimpse into its thought process as it browses the web, giving users a unique perspective on how it solves research tasks. By combining Deep Research with this new model, we expect the quality of reports to continue improving over time.