Gemini 2.0 Flash Thinking Experimental Gains New Features and Expanded Context Window
March 17, 2025Gemini is introducing an upgraded 2.0 Flash Thinking Experimental model with enhanced capabilities, including file uploads. Trained to break down prompts into step-by-step reasoning, this model now delivers responses with greater efficiency and speed. Additionally, Gemini Advanced users can now access a 1M token context window, allowing for deeper analysis and problem-solving with vast amounts of information.
Deep Research Becomes Smarter and Available for Everyone
In December, Deep Research debuted as a pioneering Gemini feature, acting as a personal AI research assistant that searches and synthesizes web information in minutes. Today, Gemini is enhancing Deep Research with 2.0 Flash Thinking Experimental, boosting its ability to assist in planning, searching, analyzing, and reporting. This upgrade enables more detailed, multi-page reports, providing deeper insights.
With Gemini’s new browsing capability, users can now see real-time thought processes as Deep Research navigates the web. This improvement ensures reports continue to evolve in quality and accuracy. Gemini is now making Deep Research available worldwide in 45+ languages, allowing anyone to try it by selecting Deep Research in the prompt bar or model drop-down. Gemini Advanced users will gain expanded access, optimizing time spent on complex research projects.
A More Personalized Gemini Experience
Gemini isn’t just a helpful AI—it’s your personal AI-powered assistant. Now, with personalization (experimental) powered by 2.0 Flash Thinking Experimental, Gemini becomes even more tailored to individual needs.
By connecting with Google apps and services, starting with Search, Gemini can deliver customized responses. For example, asking for restaurant recommendations will factor in recent food-related searches, while travel advice will reflect previously searched destinations. Users can enable Personalization via the model drop-down and retain control over their Search history, with the option to disconnect at any time.
Connected Apps Enhance Gemini’s Capabilities
Gemini seamlessly integrates with Google apps and services, and now, additional apps—such as Calendar, Notes, Tasks, and Photos—are available on 2.0 Flash Thinking Experimental. This enables Gemini to handle complex, multi-step requests with better reasoning and organization.
For instance, in a single prompt, users can ask:
“Look up an easy cookie recipe on YouTube, add the ingredients to my shopping list, and find grocery stores that are still open nearby.”
In the coming weeks, Google Photos will also be supported, allowing Gemini to analyze and retrieve information from images. Users will be able to request help with tasks like creating a travel itinerary based on past trips or finding important documents, such as a driver’s license expiration date.
Gems Now Available for Everyone
Gems, Gemini’s custom AI assistants, are now rolling out for everyone at no cost. Users can either explore pre-made Gems or create custom Gems—such as a translator, meal planner, or math coach. By accessing the Gems Manager on desktop, users can write instructions, name their Gem, and chat with it whenever needed. Additionally, file uploads can now be incorporated into custom Gems, allowing them to reference even more useful information.
With Gemini 2.0 Flash Thinking Experimental, AI-powered assistance is becoming more intuitive, personalized, and seamlessly integrated into everyday tasks.
