2025-12-05
Today at #NeurIPS2025, we present Titans, a new architecture that combines the speed of RNNs with the performance of Transformers. It uses deep neural memory to learn in real-time, effectively scaling to contexts larger than 2 million tokens. More at: https://research.google/... [image]
Google Research
Google unveils Titans, an architecture combining RNN speed with transformer accuracy for real-time learning, which can scale to context windows of 2M+ tokens
Ali Behrouz, Student Researcher, Meisam Razaviyayn, Staff Researcher, and Vahab Mirrokni, VP and Google Fellow, Google Research
2025-11-18
This Generative UI capability is built on Google's advanced foundation models & pioneering research from Google Research. This work built the core principles & first working prototype behind generative UI that unlocks the new interactive tools in both Search & the Gemini app. 📃
The Verge
Google unveils Gemini 3, its “most intelligent” and “factually accurate” model yet, with improvements across coding and reasoning, and offering less “flattery”
The flagship Gemini 3 Pro model is coming to the Gemini app and Search, with improvements across coding, reasoning, and less ‘flattery.’
AI Mode in Search dynamically creates the ideal visual layout for responses, tailored to your query. When the model detects that an interactive tool will help you better understand the topic, it uses its generative UI capabilities to code a custom simulation or tool in real time [video]
The Verge
Google unveils Gemini 3, its “most intelligent” and “factually accurate” model yet, with improvements across coding and reasoning, and offering less “flattery”
The flagship Gemini 3 Pro model is coming to the Gemini app and Search, with improvements across coding, reasoning, and less ‘flattery.’
In dynamic view, Gemini designs and codes a custom user interface in real-time, perfectly suited to your prompt. In our evaluation, raters strongly preferred dynamic view's personalized assets over standard LLM output. [video]
The Verge
Google unveils Gemini 3, its “most intelligent” and “factually accurate” model yet, with improvements across coding and reasoning, and offering less “flattery”
The flagship Gemini 3 Pro model is coming to the Gemini app and Search, with improvements across coding, reasoning, and less ‘flattery.’
Generative UI is a capability where an AI model generates dynamic, visual layouts & interactive interfaces, such as web pages, games, & apps. Today with the launch of Gemini 3, you can experience our novel implementation of generative UI in @GeminiApp & Google Search. 🧵↓ & blog [image]
The Verge
Google unveils Gemini 3, its “most intelligent” and “factually accurate” model yet, with improvements across coding and reasoning, and offering less “flattery”
The flagship Gemini 3 Pro model is coming to the Gemini app and Search, with improvements across coding, reasoning, and less ‘flattery.’
AI Mode in Search dynamically creates the ideal visual layout for responses, tailored to your query. When the model detects that an interactive tool will help you better understand the topic, it uses its generative UI capabilities to code a custom simulation or tool in real time [video]
Google Research
Google details a new implementation of generative UI that lets Gemini 3 Pro build interactive interfaces for any prompt, launching in the Gemini app and AI Mode
Google Research :
In dynamic view, Gemini designs and codes a custom user interface in real-time, perfectly suited to your prompt. In our evaluation, raters strongly preferred dynamic view's personalized assets over standard LLM output. [video]
Google Research
Google details a new implementation of generative UI that lets Gemini 3 Pro build interactive interfaces for any prompt, launching in the Gemini app and AI Mode
Google Research :
Generative UI is a capability where an AI model generates dynamic, visual layouts & interactive interfaces, such as web pages, games, & apps. Today with the launch of Gemini 3, you can experience our novel implementation of generative UI in @GeminiApp & Google Search. 🧵↓ & blog [image]
Google Research
Google details a new implementation of generative UI that lets Gemini 3 Pro build interactive interfaces for any prompt, launching in the Gemini app and AI Mode
Google Research :
This Generative UI capability is built on Google's advanced foundation models & pioneering research from Google Research. This work built the core principles & first working prototype behind generative UI that unlocks the new interactive tools in both Search & the Gemini app. 📃
Google Research
Google details a new implementation of generative UI that lets Gemini 3 Pro build interactive interfaces for any prompt, launching in the Gemini app and AI Mode
Google Research :
2025-11-08
Introducing Nested Learning: A new ML paradigm for continual learning that views models as nested optimization problems to enhance long context processing. Our proof-of-concept model, Hope, shows improved performance in language modeling. Learn more: https://research.google/... [image]
Google Research
Google introduces Nested Learning, a new ML approach for continual learning that views models as nested optimization problems to enhance long context processing
Ali Behrouz, Student Researcher, and Vahab Mirrokni, VP and Google Fellow, Google Research — We introduce Nested Learning …
2025-10-16
One of our new Gemma open models, Cell2Sentence-Scale, has identified a novel cancer therapy pathway that's been validated experimentally in living cells. Developed w/ @GoogleDeepMind & @Yale University, it looks deep into how to represent cells & biological information for AI ↓
The Keyword
Google releases Cell2Sentence-Scale 27B (C2S-Scale), a 27B-parameter foundation model for single-cell analysis built on its Gemma family of open models
We're launching a new 27 billion parameter foundation model for single-cell analysis built on the Gemma family of open models.
2025-09-14
Introducing VaultGemma, the largest open model trained from scratch with differential privacy. Read about our new research on scaling laws for differentially private language models, download the weights, & check out the technical report on the blog → https://research.google/... [image]
Google Research
Google releases VaultGemma, a 1B-parameter model it says is the largest open LLM trained from scratch with differential privacy, on Hugging Face and Kaggle
Amer Sinha, Software Engineer, and Ryan McKenna, Research Scientist, Google Research — We introduce VaultGemma …
2025-07-31
Announcing AlphaEarth Foundations, part of our Earth AI collection of geospatial models that tackle critical planetary needs: ☁️ Weather 🌊 Floods 🔥 Wildfires 🏙️ Urban planning 🏥 Public health 🚗 Urban mobility See the blog from @ymatias & Chris Phillips: [video]
Google DeepMind
Google DeepMind unveils AlphaEarth Foundations, an AI model that unifies petabytes of Earth observation data and helps scientists create detailed maps on demand
New AI model integrates petabytes of Earth observation data to generate a unified data representation that revolutionizes global mapping and monitoring