Google's AI Assistant just got a Massive Upgrade ⚡
Last week Gemini upgraded its most popular features—including Deep Research—with an updated 2.0 Flash Thinking Experimental model and personalization.
Image: Chatbot Arena Elo Score, The Keyword by Google
Gemini has become my go-to AI assistant—and there’s a reason for that—its tools are just so accessible!
I use it on my Pixel 9 (note my cognitive bias) to schedule calendar invites, create tasks, summarize emails, plan travel, and recap prior seasons of Wheel of Time every day—Wow!
Users can ground Gemini’s responses with Google Search for accuracy and personalized experiences.
I’ve also been exploring AI Studio to quickly test different models before going to Replit.
More on Gemini's new features below.
1M Context Window with File Uploads
Gemini Advanced users upload files to 2.0 Flash Thinking Experimental, with a massive 1M token context window and witness chain of thought reasoning in real-time.
Deep Research for Everyone
Research and synthesize information like never before.
Gemini now shows its thinking process, providing users with transparency and an upgraded model for insightful reports.
Everyone now has access to deep research regardless of whether you subscribe to Advanced.
Personalization Powered by your Favorite Apps
With personalization, Gemini is now connected with your Google apps (Search, Calendar, Notes, Tasks, Photos) and it can use your history to personalize responses.
Gemma 3 is a Game-changer for Agentic 👀
Now available via the Gemini API, this lightweight family of open-source LLMs can run inference on a single H100 GPU, unlocking massive potential for complex, multi-step workflows.
Agentic AI is about to become way more accessible.
DeepSeek R1 costs 32x more to run, compared to Gemma 3 27B.
