Search “DeepMind”

Google DeepMind Releases Gemini 4 Carbon Frontier Model

Google DeepMind has released Gemini 4 Carbon, calling it its strongest model, ranking first with 59 points on the Artificial Analysis Intelligence Index v5.0 and leading on SEC-bench Pro (88.2%), ExploitGym (54.8%), and CWE-bench (71.0%). The model targets long-horizon engineering tasks; an internal team once had it run continuously for several days to investigate and fix hidden defects in a large codebase, and it rose 89 elo to 4th place (1768) on WebDev Arena. Starting this week, Carbon will first be opened to some cybersecurity defenders, then gradually expanded through the Fairwind Program, and only afterward made available to developers, enterprises, and consumers; after the introductory period, pricing will be $4 per 1M input tokens and $20 per 1M output tokens.

Google DeepMind Releases Gemini 4 Carbon, Focused on Cybersecurity and Long-Horizon Task Capabilities

Google DeepMind has released Gemini 4 Carbon, calling it its strongest model, ranking first with 59 points on the Artificial Analysis Intelligence Index v5.0, and achieving 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench. The model is aimed at long-horizon engineering tasks; an internal team once had it run continuously for several days to investigate and fix hidden defects in a large codebase, and it rose to 4th place on WebDev Arena with 1,768 points. Starting this week, Carbon is first being made available to select cybersecurity defenders, then gradually expanding through the Fairwind Program, and only afterward to developers, enterprises, and consumers; after the introductory period, pricing is $4 per 1 million input tokens and $20 per 1 million output tokens.

Paper from Google DeepMind and other institutions: the best AI agent solved 49 of 50 unpublished research-level physics problems

A new paper from Google DeepMind and several top universities shows that the best-performing AI agent in the test solved 49 of 50 unpublished research-level theoretical physics problems, each roughly equivalent to a day's work for a domain expert, at a cost of under $3 per problem. The problems were contributed by experts from their own research, and the agents answered independently, split into two groups—with and without internet access—and were scored only on the final numerical answer. GPT-6-Astra, with internet access in Codex, solved 49; it and Opus 5.5, without internet access in Claude Code, each solved 46, and removing internet access changed accuracy very little, leading the authors to suggest that the models may have memorized a large amount of physics literature.

Google DeepMind paper: Proactive AI agents create stress in some teams

A new Google DeepMind paper finds that the same AI assistant was seen by some employees as a savior and by others as a source of stress, because it acts on its own without being asked. The agent ran for 5 months across more than 20 teams at a large tech company, and researchers interviewed 17 users; it treated outdated documents as current, tagged a VP on 16 bugs, shared a poster draft before the author was ready, and misread sarcastic tone. Some users called it a savior, while others felt surveilled and moved sensitive discussions to chats without the agent; teams that actively chose the agent forgave its early mistakes, while those who felt it was imposed on them did not. The authors recommend obtaining team consent when deploying agents, separating permissions, and providing an easy way to undo actions.

DeepMind has four Gemini agents team up to defeat the Ender Dragon in Minecraft

Google DeepMind had four Gemini 4 Argon agents team up to play Minecraft from scratch, defeating the Ender Dragon after 27 hours of real time, during which they died 605 times, wrote more than 2,700 programs, and produced a total of 63,300 lines of code. The four agents had no human-written programs to rely on; chopping trees, mining, and crafting equipment all required them to write and test their own code, and they also shared experience with each other and reported the locations where teammates died. The article also reviews work from the 2016 Malmo platform and the 2019 MineRL diamond-mining competition to 2022's OpenAI VPT and 2023's Voyager, among other efforts that use Minecraft to study AI.

The UK will establish the world's first AI biosecurity laboratory, supported jointly by OpenAI, Anthropic and Google DeepMind

The UK announced it will establish a national-level AI biosecurity laboratory, supported jointly by OpenAI, Anthropic and Google DeepMind, with the government investing £100 million for this purpose, while OpenAI and Anthropic will additionally provide AI model usage credits worth up to £200 million. The laboratory is officially described as the world's first institution dedicated to AI biosecurity; it will use AI to detect and monitor biosecurity threats and will cooperate with Google DeepMind and life sciences company Basecamp Research, though its specific functional structure has not yet been fully disclosed. This package of measures also includes an additional £120 million for the UK AI Safety Institute and an additional £23 million for the planned UK AI Economy Institute.

UK to establish the world's first AI biosecurity laboratory, jointly supported by OpenAI, Anthropic and Google DeepMind

The UK Cabinet Office announced that the UK will establish a national-level artificial intelligence biosecurity laboratory for detecting and monitoring biosecurity threats, jointly supported by OpenAI, Anthropic and Google DeepMind. The UK government will invest 100 million pounds for this purpose, and OpenAI and Anthropic will additionally provide AI model usage credits worth up to 200 million pounds; the laboratory will also cooperate with Google DeepMind and life sciences company Basecamp Research. The announcement also includes an additional 120 million pounds for the UK AI Safety Institute and an additional 23 million pounds for the UK AI Economy Institute under preparation.

Pichai responds to Hassabis stepping down as DeepMind head and Jeff Dean leaving Google

Alex Heath relays Sundar Pichai's explanation for Demis Hassabis stepping down as head of DeepMind and Jeff Dean leaving Google. Pichai called it a natural evolution after the individuals' long-term commitment, and mentioned that Jeff Dean will go work on Discovery Loop; he said frontier model development has become high-intensity, large-scale work spanning data, compute, algorithms, and team management, while Hassabis wants to devote more time to safely advancing the frontier and using AGI to solve other problems, with AlphaFold and Isomorphic as examples of such applications.

Google DeepMind beats Minecraft with four Gemini 4 Argon agents

Google DeepMind published the paper "A team of Gemini 4 Argon agents beats Minecraft": four Gemini 4 Argon coding agents autonomously divided labor in a randomly generated Minecraft survival world, accumulating 605 deaths and writing 2,783 programs, defeating the Ender Dragon in 27 hours and 14 minutes. The experiment used Minecraft Java Edition 1.20.4, Easy difficulty, a random map, and no cheats; the agents read the environment via Mineflayer and wrote JavaScript to command characters, with roughly 24,000 model calls and 1.35 billion input tokens over the whole run, 91% of which came from context caching. The authors note that about 22% of game programs contained errors and the server restarted 6 times; the paper did not demonstrate stable completion through repeated tests on multiple random maps, and at discounted API prices the completion cost is estimated at about $520.

Google DeepMind, together with WRI and IIASA, releases ForTy 2020, a 10-meter-resolution global forest type map

Google DeepMind and Google Research, together with the World Resources Institute (WRI) and the International Institute for Applied Systems Analysis (IIASA), released Forest Typology (ForTy) 2020, which classifies every 10×10-meter parcel of land worldwide into six forest and land-use types, serving as a key baseline for the European Union's deforestation regulation. The team trained the model on about 1.7 million labeled locations and used AlphaEarth Foundations embedding vectors as a key input; in validation at more than 9,000 randomly sampled points, the forest type classification accuracy was 90%, and the accuracy in distinguishing forest from non-forest was 95%. A companion year-by-year time series from 2017 to 2025 is also openly available and can be used free of charge under an open license on Google Earth Engine.

Google DeepMind and Cambridge paper: AI agent autonomously discovers exact rules for 11 unsolved Turing machines

A new paper from Google DeepMind and Cambridge shows that a Gemini-based agent can autonomously discover new mathematics, finding and providing exact rules for 11 Busy Beaver Turing machines that previously had no analysis on record. The agent tests edge cases in a simulator, drafts informal proofs, and then completes the proofs in the Lean proof checker; all the relevant rules passed Lean verification. The paper states that this reflects AI moving from answering mathematical questions posed by humans toward open-ended discovery.

English translation is being prepared.

Former Google DeepMind researcher founds London AI company Metis Reasoning, focusing on compute-efficient reasoning

Metis Reasoning, a London AI company co-founded by former Google DeepMind research lead and AlphaGo co-creator Thore Graepel, has officially launched, seeking a compute-efficient path distinct from the frontier models of Anthropic and OpenAI, and is believed to have raised tens of millions of dollars in initial funding. The company has a team of 12 people, including researchers, mathematicians, robotics experts, and technology entrepreneurs. It believes that LLMs have fundamental limitations in reasoning, that results are difficult to verify, and that there is a lack of audit trails; its approach is more verifiable, more robust, and safer, and it says it has already achieved some positive results on frontier reasoning problems at a cost far below that of frontier models. Metis has not yet announced a release date for its first AI model.

DeepMind AlphaProof Nexus appears in《Science》: autonomously solves 9 Erdős problems

Google DeepMind's automated mathematical proof system AlphaProof Nexus appeared in the October 8 issue of《Science》, autonomously solving 9 of 353 open Erdős problems, two of which had remained open since 1970. The system uses Gemini 3.1 Pro to draft proofs in Lean, with a compiler verifying line by line and feeding back iterations; the reasoning cost per problem is several hundred dollars. The paper's ablation results show that a simplified system that removes complex structures such as the sub-agent tournament and retains only the large model plus Lean feedback can also solve these 9 problems, though the hardest few are more expensive.

English translation is being prepared.

English translation is being prepared.

Google DeepMind launches general physical reasoning model Gemini 4 Argon

Google DeepMind officially launched the general physical reasoning model Gemini 4 Argon and released a demo video of its collaboration with the Apptronik Apollo humanoid robot. The robot only watched a video demonstrating beanbag tossing, practiced repeatedly, and ultimately made 7 consecutive successful throws. The model possesses visual semantic reasoning, physical understanding, and code generation capabilities, and can be used for robot control, simulation environment construction, physical data analysis, and motion optimization. In pure reasoning mode, it scored 87% on the ERQA benchmark. The demo also covers generating MuJoCo simulation environments from robot-perspective videos, human-perspective videos, and hand-drawn sketches, as well as using simulation data to raise the real robot's grasping success rate from 56% to 75%.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.