Every brief from the last seven days
31 briefs since Aug 29, grouped by desk. The same list goes out as the weekly email; sign up in the footer to get it delivered.
Intelligence3 briefs
A benchmark firm measured what agents cost in energy and found accuracy and footprint have come apart
Vals AI ran 16 open models through long tasks with an emissions estimator. The most accurate model used about 12 times the energy of the runner up for four more points. Token price predicted none of it.
Nvidia says an open Nemotron model outscored the top human at IOI 2026
Ultra-CC posted 535.4 of 600 against a best human 498.27, live, under contest rules. The score is real. The claim that it generalizes is not yet.
Researchers turned 1,000 GitHub repos into 5,000 agent skills and more than doubled a fixed model’s MLE-bench score
With the same GPT-5.5 backbone and the same compute, a library of distilled procedures lifted a coding agent from 31 to 73 percent. The knowledge was sitting in READMEs the whole time.
Frontier9 briefs
India puts its first operational imaging satellite in geosynchronous orbit, five years after losing the last one
GSLV-F17 lofted the 2,367 kilogram EOS-05 to a higher orbit than planned. The rocket that failed in 2021 has now carried its heaviest payload, and India gets a camera that never leaves the subcontinent.
Ray Kurzweil joins a startup that wants to put a brain interface up your nose
Subsense has 27 million dollars, mouse studies, and a plan to reach neurons with inhaled nanoparticles and a magnetic cap. The name it just added is worth more than the money, and the science is still preclinical.
Abu Dhabi releases six models under Apache 2.0 and calls them fully open, with caveats
MBZUAI’s K2 Horizon spans 0.9 to 375 billion parameters with weights, code and data promised for all. For the two largest, the data and code are still to come.
OpenAI ships a model it rates Critical for cyber, and admits it is harder to watch
GPT-6 Astra saturates the offensive security benchmark OpenAI uses to measure danger, and arrives with a disclosure that its reasoning is less legible than the model it replaces.
Microsoft AI prices transcription at 10 cents an hour and aims it at OpenAI
MAI-Transcribe-2 claims first place on FLEURS across 60 languages and a 72 percent price cut in five months. The introductory price expires with the year, and the real rate is unknown.
Quantum Motion adds investors on the bet that a quantum computer should cost what a chip fab already makes
The London company claims silicon spin qubits on standard CMOS can cut machine cost a hundredfold. Its new backers, a Maryland lab and a DARPA slot are the evidence it has to show before 2028.
Saudi Arabia’s national Arabic model is a 428 billion parameter MiniMax build
HUMAIN says humain-m3 beats GPT-5.6 and Opus 5 on seven Arabic benchmarks. The weights come from a Chinese lab, the scores come from HUMAIN, and the release is a month away.
Gemini 3.8 Flash ties Opus 5 on coding benchmarks at a price that doubles on 1 January
Google’s fourth Flash model in four months is a harder working 3.7 Flash, not a new base. It burns 30 percent more tokens per task, and the discount that makes it cheap expires with the year.
Anthropic ships Fable 5.1 and cuts cache reads 75 percent, the price lever agents actually feel
Base prices are unchanged at 10 and 50 dollars. The cut lands on the tokens an agent reads over and over, which is why Anthropic can claim 45 percent savings on agentic work without touching the headline rate.
Systems6 briefs
Microsoft’s Project Zenith is a Windows for developers that runs 30 billion parameter models without a meter
The requirement is 64 gigabytes of unified memory and 250 gigabytes a second of bandwidth, first on AMD’s Ryzen AI Halo. The pitch is a local coding model that costs nothing per token, which is a pitch against the cloud Microsoft also sells.
Trezor says a shipping vendor kept five years of data it certified as deleted, exposing 67,000 more
The hardware wallet maker had written assurances that ShipMonk had erased old orders. A Metabase zero day proved it had not. Total exposed now exceeds 80,000 people who own crypto.
DeepSeek orders 160,000 Huawei chips for a gigawatt inference site in Inner Mongolia
The order moves China’s best known lab onto domestic silicon for serving, not training. Huawei’s memory supply, not DeepSeek’s ambition, sets the timeline.
Tenable and OpenAI will inspect community built agents before enterprises run them
A registry of 100 plus AI agents, skills and MCP servers gets a security review layer built on GPT cyber models and human researchers. It is the first attempt at an app store gate for the agent supply chain.
Figure commits 3.5 billion dollars of compute it has not raised to Nscale
The humanoid robot maker signs for up to 100,000 Vera Rubin GPUs in Texas from 2027, with the cloud provider taking equity in return. Robotics now needs a frontier lab’s infrastructure bill.
Broadcom just told the market the accelerator business doubles twice, and that supply is the only thing in the way
A $16.7 billion quarter came with a three year ladder to $230 billion, and the binding constraint management named was power, wafers and memory rather than orders.
Markets7 briefs
Gimlet Labs raises 300 million dollars to split inference across rival chips
Arm and Microsoft joined a16z in a round that values a chip agnostic inference cloud at 3 billion dollars, up from about 400 million six months ago. The bet is that the CUDA lock does not hold at inference.
Crusoe triples to 30 billion dollars on a 13 billion dollar Jane Street contract
A gas flare crypto miner turned data center builder raised 3 billion dollars at three times its October valuation. The customer that made the round possible is a trading firm, not a lab.
Adobe hands the CEO job to its enterprise chief as AI design rivals close in
Anil Chakravarthy, who ran the customer experience business and global sales, takes over from Shantanu Narayen on 1 December. The choice tells you which half of Adobe the board thinks is defensible.
Ultrahuman raises 70 million dollars to turn a health ring into a computer, with Qualcomm inside
The Bengaluru company tripled its valuation to 365 million dollars on a 140 million dollar run rate and a fight with Oura it survived. The new money is for a ring that runs software, which is a different product from the one that made it.
Nvidia buys Hugging Face for 12.9 billion dollars and promises to keep it open
The chipmaker is paying about 86 times annualized revenue for the place where open models live. The filing says the platform stays neutral. The incentives say watch closely.
Cognition heads for a 47 billion dollar mark three months after 26 billion
The Devin maker is fielding close to 10 billion dollars of interest for a 1 billion dollar round, and the repricing says more about the coding agent market than about Cognition.
South Korea shipped 46.65 billion dollars of chips in a month, half of everything it exported
Semiconductor exports tripled in a year on AI memory demand and now carry the national trade balance. That is a boom for Samsung and SK Hynix and a concentration risk for the country.
Policy6 briefs
Europe’s defence ministries push back on the cloud sovereignty law meant to protect them
The Cloud and AI Development Act reserves the top tier for European owned providers. The workloads that would sit there, F-35 systems and NATO coordination, run on American clouds by design, and the generals want to keep them there.
NHTSA opens an audit of 1,000 Cybercabs hours after Tesla put them on Austin streets
Tesla self certified a car with no steering wheel, pedals or mirrors. The regulator wants to see the reasoning, and the rules it would need to have followed were written for humans.
Sanders and Casar propose a permanent ban on superintelligence with 20 year sentences
The bill would pause frontier development until a new federal regulator writes safety rules. It has no text, no number and no cosponsors yet, and it is aimed at a fight rather than a vote.
Lutnick confirms phase two chip tariffs: build in America or pay to enter
The Commerce secretary says the next Section 232 round will reach servers and laptops and tie relief to US fab investment. Rates, exemptions and timing are still unspoken.
China removes 5.61 million AI generated items and penalizes 49,000 accounts in four months
The second phase of the Qinglang campaign turned labeling rules into enforcement at platform scale. It is the largest AI content action any government has reported.
California passed an AI code in one session, and handed the whole thing to one desk
The legislature adjourned on 31 August with roughly 30 AI bills cleared, an audit registry, a private right of action over chatbots and a 30 September veto deadline that now sets national compliance.