Latest

Unisound's U2-Flash builds a working agent in 18 minutes

The Frontier

Unisound's U2-Flash builds a working agent in 18 minutes

Two stories from China's model race this morning, one about the model and one about the machine it runs on — both arguing the same thing: the frontier metric has moved from raw size to cost per finished task. Unisound released U2-Flash, a "flash"-class model it says handles mainline agent, coding and office work at roughly 10 billion active parameters. The company claims under 4% of total parameters fire on any given call, so inference cost anchors to the activated slice rather than the whole c

BYD's new cockpit agent runs on Alibaba's Qwen

The Arena

BYD's new cockpit agent runs on Alibaba's Qwen

China's model race keeps moving from leaderboards into hardware people actually buy: a cockpit agent and a same-week model update — while Scott Aaronson reports the labs may now be holding back what their models solve. BYD's next-generation smart cockpit runs on Alibaba's Qwen, the first time China's most-deployed open model family moves into the car as a commercial product rather than a hobbyist port. At a September 14 launch event where BYD introduced its "Didixia" super agent and the Denza N

Nvidia Earth-2 turns UK air-pollution forecasting into a desktop job

The Frontier

Nvidia Earth-2 turns UK air-pollution forecasting into a desktop job

A generative weather model gets retargeted at smog — and China's Doubao refreshes its flagship API with agent delivery as the pitch. University of Manchester physicists trained NVIDIA's Earth-2 CorrDiff — a generative downscaling model built for weather — into a UK-wide air pollution model in two days, and the same workflow now runs inference on a desk-side DGX Spark. David Topping's group fed the model a year of hourly UK pollution fields simulated on Isambard-AI, the Bristol supercomputer, on

Zuckerberg rejects a coordinated AI pause, calls alignment a moat

The Guardrails

Zuckerberg rejects a coordinated AI pause, calls alignment a moat

The pacing fight that has owned the last two weeks got its fullest industry-side answer — from the company that has spent months proving it can wait. Meta CEO Mark Zuckerberg published the industry's most detailed rebuttal to a coordinated AI slowdown, arguing that market pressure, not a labs' pact, will police frontier models. In a post on X on Tuesday, Zuckerberg wrote that "every lab has the responsibility and incentive to move at the pace required to train its models safely," and that "trus

Vidu S2 rewrites the video while it is still playing

The Frontier

Vidu S2 rewrites the video while it is still playing

The two stories nobody outside China is covering this morning are both about the same thing: what happens to AI video and AI serving once the model has to answer in real time. Shengshu Tech made the live video stream editable while it plays. Vidu S2, released September 16, is two models — Avatar for live interactive digital humans and Editing for changing a running video — and the avatar's live output moved from 540p to 720p while the frame rate holds, so a viewer's request to change someone's

vivo ships four Blue LM models aimed at phones that actually finish tasks

The Arena

vivo ships four Blue LM models aimed at phones that actually finish tasks

The quiet part of the model race is happening on handsets, and vivo just laid out its full hand at its 2026 developer conference. Two of today's other stories point the same way: memory, not compute, is the industry's binding constraint, and the biggest brain map ever built is now a playground. vivo rebuilt its Blue LM line around agents that act, not chat. At the company's 2026 developer conference on September 16, vivo announced four core models — Blue LM-RealTime, Blue LM-Nano, Blue LM-Flash

Huawei fixes its network agent's 'topology amnesia' with a live map

The Frontier

Huawei fixes its network agent's 'topology amnesia' with a live map

Two stories out of China's phone-and-carrier world this morning, and both make the same bet: the next agent gains come from the harness around the model, not the model itself. Huawei's GTS AI team says it cured its troubleshooting agent of "Topology Amnesia" by giving it a map that updates as it probes. The diagnosis is sharp: a text-only agent can parse any single CLI output fine, but deep into a multi-device, multi-path network with firewalls and ECMP in the mix, the local observations pile u

Instinct wants a $10B valuation weeks after a $2.5B round

The Arena

Instinct wants a $10B valuation weeks after a $2.5B round

Two stories about what the AI race is actually buying: one company raising four times its own valuation to cover a compute bill it can't pass to users, and one chip vendor trying to make that bill smaller. Plus a paper showing what structural search finds when it stops reading sequences. Instinct is negotiating to raise as much as $1 billion at a valuation of about $10 billion, according to The Information, with Sequoia and Benchmark in talks to lead a round that is not yet final and may still

Workiva ships no-code agents for work that has to survive an audit

The Stack

Workiva ships no-code agents for work that has to survive an audit

Enterprise AI's next fight is not capability, it is defensibility. Workiva spent its annual conference selling the version of agents that ends in a signature. Workiva put its agent layer in customers' hands at Amplify 2026, unveiling Agent Studio — a no-code capability for building, customizing and deploying AI agents inside its own reporting platform — alongside three agentic products for regulatory filings and a fourth for internal audit testing. The pitch is not a smarter model, it is a gove

Trump says he'll meet AI chiefs on a government stake

The Guardrails

Trump says he'll meet AI chiefs on a government stake

Washington moved on the AI industry's cap table, its power bill and its chain of command in a single day — three levers, one direction. President Trump said he expects to meet the leaders of major US AI companies at the White House, likely next week, to discuss the government taking a financial stake in their businesses. Speaking to reporters aboard Air Force One, Trump framed the idea as profit-sharing rather than nationalization: the aim is to "create almost a partnership with the American pu

Huawei's chairman says its computing goal is to become Nvidia

The Arena

Huawei's chairman says its computing goal is to become Nvidia

Two infrastructure stories today: the ambition inside Huawei's compute business finally goes on the record, and a security vendor puts a hard number on whether detection rules actually fire. Huawei's supervisory board chairman Guo Ping told new employees that the company's target in ICT and computing is to become the Nvidia of the field — and that its real advantage is not owning a frontier model, but running everyone else's. The remarks come from roughly 10,000 characters of Q&A minutes that H

Factory raises $200M at a $5B valuation, tripling in five months

The Arena

Factory raises $200M at a $5B valuation, tripling in five months

Two stories today: the enterprise coding-agent market repriced again, and South Korea's chip giants started signing students before they graduate. Factory raised $200 million at a $5 billion valuation, more than tripling the $1.5 billion mark it took in April — a five-month reprice that says enterprise buyers, not consumers, are setting the pace in AI software. The round was backed by Blackstone, Khosla Ventures, Sequoia Capital, Insight Partners, Evantic Capital, Sound Ventures, NEA, Mantis VC

See all