Google DeepMind

Gemini for Science is a new suite of experimental AI tools and agentic prototypes that help researchers generate hypotheses, run large — scale computational experiments, and synthesize scientific literature.

Gemini 3.5 Flash, released May 19, 2026, pairs frontier — level reasoning with agentic capabilities to plan, execute and iterate on multi — step real-world workflows and is available now across consumer, developer

Project Genie can now ground its generative world model in Street View imagery for U.S.

Calico Life Sciences used DeepMind’s multi‑agent system Co‑Scientist to synthesize fragmented aging literature and produce a novel, testable hypothesis that the integrated stress response (ISR) is regulated by metabolic

In May 2026 Professor Clare Bryant at the University of Cambridge used the multi‑agent AI Co — Scientist to generate and rank hypotheses from a grant summary and a full proposal, which led the team to prioritize

In a May 19, 2026 research post, Omar Abudayyeh and Jonathan Gootenberg reported that their lab used a multi‑agent AI called Co — Scientist to generate and validate genetic leads that push human cells away

An AI multi‑agent assistant, Co — Scientist, compressed months of ALS literature to help MIT mechanical engineer Ritu Raman and Boston Children’s chemical biologist Ryan Flynn convert ideas into ranked, testable

On May 20, 2026, an AI research lab and the Singapore Government unveiled a National Partnerships for AI program to deploy frontier models across healthcare, scientific discovery and education while training local

WeatherNext produced early, high‑confidence forecasts that Hurricane Melissa would rapidly intensify and make Category 5 landfall in Jamaica in October 2025, giving forecasters extra lead time to warn communities

On May 19, 2026, DeepMind published a Nature paper introducing Co‑Scientist, a Gemini‑based multi‑agent system that generates, debates and refines scientific hypotheses;

Introduced May 20, 2026, the Running Guide agent is an on‑device accessibility system that pairs low‑latency edge segmentation on a chest‑mounted Pixel 10 Pro with Gemma 4 E4B multimodal reasoning to deliver real‑time

Google DeepMind has launched its "AI co-clinician" research initiative, aiming to integrate advanced AI into clinical practice to amplify physician expertise and deliver higher quality patient care.

Google DeepMind unveiled Gemini 3.1 Flash Live—the latest generative model optimized for native real-time audio processing.

Google DeepMind unveiled Decoupled DiLoCo — an architecture for distributed LLM training that isolates failures, reduces network requirements, and enables training models across remote data centers.

Google DeepMind joins forces with Accenture, Bain & Company, BCG, Deloitte, and McKinsey to accelerate the digital transformation of enterprises worldwide through the large-scale adoption of advanced AI technologies and to overcome the existing gap in their adaptation.

The updated version of Gemini Robotics ER 1.6 offers enhanced spatial perception and multilayered understanding, opening new possibilities for autonomous robots.