Thursday, 29 October 2026
Google's Gemini 4 Carbon Goes to Cyber Defenders First
Google’s Gemini 4 Carbon Goes to Cyber Defenders First
Google DeepMind announced Gemini 4 Carbon on Wednesday, calling it “our strongest model with state-of-the-art cybersecurity and long horizon capabilities.” The headline numbers are Google’s own: first place on the Artificial Analysis Intelligence Index v5.0 with 59 points, 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploiting them, and twice Argon’s success rate at writing secure code. The cyber scores, Google notes, were measured without guardrails. That is why ordinary users can’t have it yet: this week it goes only to a small group of cyber defenders, then to trusted partners, and to developers, businesses and consumers “as soon as possible,” once the safeguards are refined. Google is also putting it through the US government’s voluntary pre-release testing. Introductory API pricing is $2 per million input tokens and $10 per million output, doubling later. A model good enough at breaking in that the first customers are the people paid to keep others out.
Source: Google, October 28, 2026
Google Cloud Revenue More Than Doubles to $30.7 Billion
Alphabet reported its third quarter on Wednesday, and the AI bill finally shows up on the income side too. Google Cloud brought in $30.7 billion, up 103% from a year earlier. Total revenue was $127.6 billion, up 25%; Search and other grew 12% to $63.4 billion, and operating income rose 30% to $40.6 billion. Sundar Pichai’s line in the release: “Our AI investments are driving overall Search query growth, including commercial queries” — the chatbot that was supposed to kill search, apparently feeding it. The spending side kept pace: Alphabet bought $51.4 billion of property and equipment in the quarter, mostly the data centres the cloud sells.
Microsoft’s Azure Grows 48%, With $677 Billion in Contracts Still to Deliver
The same afternoon, Microsoft reported its fiscal first quarter: revenue $92.3 billion, up 19%; Microsoft Cloud $63.9 billion, up 30%; Azure up 48%; net income $36.6 billion. The number that matters most is the queue: commercial remaining performance obligation — contracts signed but not yet delivered — rose 73% to $677 billion. Satya Nadella added figures of his own on X: daily Copilot users up nearly 50% in a quarter after Chat, Cowork, Autopilot and Code were merged into one Copilot, and more than 50 million agents running in Agent 365. Those are company claims; the filing doesn’t break them out. Two giants, one evening, and both say the same thing: customers aren’t trying AI any more, they’re signing for years of it.
OpenAI Says an Unreleased Model Disproved the Exponential Time Hypothesis
OpenAI said on Monday that an internal model produced a refutation of the Exponential Time Hypothesis — the 1999 conjecture that some hard problems, such as 3-SAT, cannot be solved much faster than by trying everything. The model found the argument on 16 October, and a version checked in the Lean proof assistant followed a day later. OpenAI also claims results in harmonic analysis (the local smoothing and Kakeya conjectures) and in physics, a proof that four-dimensional Yang-Mills theory exists. None of it is verified yet. The hypothesis’s own authors, Russell Impagliazzo and Ramamohan Paturi, with MIT’s Ryan Williams, have agreed to examine the claim and wrote: “Our involvement should not be taken as certification of the claim.” The physics group, led by Nobel laureate David Gross, is holding the AI’s material back and writing it up itself. It does not mean P equals NP. It may mean the people who proposed a conjecture now spend a year checking a machine’s homework against it.
Source: The Next Web, October 26, 2026