Congratulations to the authors of "AfriMed-QA: A Pan-African, Multi-Specialty, Medical Question-Answering Benchmark Dataset", recipient of the Best Social Impact Award at #ACL2025! aclanthology.org/2025.acl-long.…

Google's AI models for flood forecasting and cyclone prediction will be deployed in Vietnam. Our partnership with Weathernews Inc (WNI) will operationalize these tools to VNMHA, which aims to enhance disaster preparedness and community safety. Read more: global.weathernews.com/news/18480/

We’re bringing a version of Deep Think that achieved gold-medal status at IMO to Ultra subscribers in the @Geminiapp (+ the official version is now in the hands of mathematicians).  Toggle it on when reasoning through complex scientific literature, tackling a coding problem that

See mathematician Michel van Garrel talking about how our latest Gemini Deep Think model was able to prove a conjecture using a very different approach than he was considering.

NotebookLM is now available for younger users to help them understand their class materials. Here's our attempt to translate this news for them: "Need to lock in for the upcoming school year? NBLM is goated 🐐! Turn your notes into podcasts, study guides, and mind maps, no cap!"

Providing advanced AI tools and training to students is an important step for education. It gives the next generation the resources to build a deeper understanding of new concepts and tackle complex problems. Google is now offering our most advanced Gemini AI tools and training

New research on AMIE proposes a path for responsible AI integration in healthcare. This model demonstrates how AI can improve clinical decisions while ensuring human accountability. More in today’s blog: research.google/blog/enabling-… Research paper: arxiv.org/abs/2507.15743

New paper in Nature Medicine introduces PH-LLM, a Gemini-based model for personalized health. It integrates wearable device data to provide insights and coaching for sleep and fitness. The model exceeded human expert scores on professional exams and performed on par with experts

Grateful for the music and the inspiration of Oscar Peterson. I was fortunate to be in one of his last performances and to not only experience his wonderful music but also his warmth and generosity while having a conversation at the back stage.

100M videos created in FlowbyGoogle - our AI for filmmakers - since May!  Huge milestone, and now Ultra Subscribers get 2X the AI credits to keep the creativity flowing - labs.google/flow

Google is sharing a technical paper on a comprehensive methodology for measuring the energy, emissions, and water impact of Gemini prompts. Over a 12-month period, the energy consumption per median Gemini Apps text prompt decreased by 33x and the carbon footprint by 44x. This is

Build, ship, repeat 🚀Google Search is: → Expanding AI Mode to 180+ new countries & territories in English And in AI Mode in the U.S., we’re: → Rolling out agentic web browsing capabilities → Launching personalized recommendations → Enabling sharing of AI Mode responses

Our new research introduces MAD and MAD2R, scalable algorithms for differentially private partition selection. The work addresses the challenge of analyzing large datasets while preserving user-level privacy.

Veo 3 is now free for the weekend for everyone to try  - can’t wait to see what you’ll create! 🎬

Good idea. Let’s double it for everyone. This weekend only, ending tomorrow (8/24) at 10pm PT: 👉 Gemini free users, you now have up to 6 free video generations for the weekend. 👉 Pro users, your pack has doubled to 6 per day for the weekend.  👉 Ultra users, your pack has

Introducing EmbeddingGemma, our newest open model that can run completely on-device. It's the top model under 500M parameters on the MTEB benchmark and comparable to models nearly 2x its size – enabling state-of-the-art embeddings for search, retrieval + more.

A preprint for a new AI system built with Gemini helps scientists write expert-level empirical software. It generates research ideas, implements them as executable code, and uses tree search to iteratively improve solutions. - We tested it on six challenging and diverse