This week, super intelligence learned to build worlds you can walk through, Europe opened a trillion parameter model to everyone, and Canada decided to own a piece of its own champion.

Google DeepMind taught a model to build worlds, not just pictures. On October 27, DeepMind unveiled Genie 4, a world model that generates interactive, photorealistic environments you can navigate and modify in real time from a description or an image. Unlike a video generator, Genie 4 simulates what happens next as you move, and it now adds sound, including ambient audio and dialogue. Google is open about the limits, including physical consistency and longer interactions, but the direction is clear.

Why it matters for how we live and build: a world you can generate is a place you can rehearse in, whether that is a game, a training ground for robots, or a design you want to walk through before you pay to build the real one.

Mistral opened a trillion parameter model to the world. On October 29, Mistral released the weights of Large 4, a 1 trillion parameter multimodal model trained in Europe, under a modified MIT licence that lets anyone use, modify and sell with it, while companies offering it as a hosted service buy a separate licence. Mistral prices its own hosted version at $0.70 per million input tokens and reports coding agent results effectively tied with the leading open weight rival while using far fewer tokens per task.

Why it matters for how we live and build: open weights mean a serious model can live on infrastructure you control, with your data staying home, and that changes the bargaining power of every builder who rents intelligence by the token today.

Google kept its strongest model behind a door, on purpose. On October 28, Google announced Gemini 4 Carbon, which it says ranks first on version 5.0 of the Artificial Analysis Intelligence Index at 59, with strong reported results at finding and patching software vulnerabilities. Access starts with a subset of cyber defenders through its Fairwind Program while the model goes through the United States government voluntary pre-release testing, with no date yet for wider release.

Why it matters for how we live and build: the same skill that patches a flaw can be used to exploit one, so staged access is becoming a design choice in itself, and who gets through the door first is now part of the product.

Attackers, meanwhile, are not waiting for permission. CrowdStrike reported on October 29 that a group it tracks as VAPOR PANDA ran an AI agent framework called ARTEX against more than 200 organisations in at least 14 countries over 11 days from October 17, reaching deeper access at nine of them. CrowdStrike says the agents ran mainly on a DeepSeek model with a human operator directing them, an escalation from one intrusion at a time to several at once.

Why it matters for how we live and build: defence has to assume tireless, parallel attackers, which is exactly why the gated defender models above matter, and why audit trails and access limits are no longer back office details.

Anthropic stepped into the public light. On October 26, Anthropic filed its public Form S-1 to list Class A shares on the Nasdaq under the proposed ticker ANT, with no share count or price range set yet. The filing reports $16.3 billion in revenue for the first half of 2026, up from $1.2 billion a year earlier, alongside very large future compute commitments and a governance structure that keeps voting control with its founders.

Why it matters for how we live and build: public filings turn private momentum into public numbers, and for the first time anyone can read, in one document, what building frontier super intelligence actually costs.

Canada bought a seat at its own table. On October 29, the Government of Canada announced an equity investment of up to about C$1.6 billion in Cohere, in stages, as part of a Series E round led by Germany’s Schwarz Group. In return, Cohere keeps its Canadian incorporation and headquarters and commits at least C$1.7 billion to additional computing infrastructure in Canada by 2030, with Nvidia and Bell among the investors named.

Why it matters for how we live and build: sovereign super intelligence is becoming industrial policy, and countries that own a stake in their champions keep the talent, the infrastructure and the decisions closer to home.

OpenAI moved deeper into the working day. OpenAI and ServiceNow announced an expanded partnership that connects business workflows directly with ChatGPT and Codex, covering work such as investigating IT problems, updating service records, helping developers build applications and supporting cybersecurity operations. Further capabilities are scheduled to arrive through 2027, and ServiceNow plans to give its own employees ChatGPT Work and Codex.

Why it matters for how we live and build: the contest is shifting from answering questions to completing tasks inside the systems where work already lives, which makes permissions and oversight the real product decisions.

Robots got a way to count useful work. At its Innovation Day in Shanghai on October 29, AGIBOT introduced Worken, a proposed framework that measures a robot fleet by active robots multiplied by autonomous working hours and an output efficiency score against a human reference standard. The company also launched its G2 Max industrial robot and Link-U OS 1.0, which it positions as an operating system built for embodied intelligence.

Why it matters for how we live and build: industries scale what they can measure, and a common measure of useful robot work turns impressive demonstrations into something a business can actually budget for.