Newsletter

China AI

Deep Dive — DeepSeek V4.1-Flash-1017: 77% on DeepSWE for 20 cents

The Frontier

Deep Dive — DeepSeek V4.1-Flash-1017: 77% on DeepSWE for 20 cents

DeepSeek refreshed its cheapest model on Friday and the headline number is the one buyers actually care about: 77.0 percent resolution on DeepSWE v1.1, per the model card, at just under $0.20 of compute per task on the company's cost chart. That chart puts Claude Opus 5, GPT-6 Astra and Gemini 3.8 Flash somewhere between $2 and more than $10 for the same benchmark, and none of them above the small Chinese model's score — a gap The Decoder summarized as beating larger rivals on agentic tasks…

Taiwan makes AI chip smuggling to China a criminal offense

The Guardrails

Taiwan makes AI chip smuggling to China a criminal offense

Taipei is turning the black market in advanced AI hardware into a crime it can actually prosecute — and a researcher says the field has spent a decade misreading how its own models work. Taiwan's government plans to add AI servers and cutting-edge semiconductors to its list of controlled exports, and once the measures take effect, shipping restricted AI hardware to China becomes a criminal offense under the Foreign Trade Act, Minister of Economic Affairs Kung Ming-hsin told Bloomberg News.

JD.com runs its 11.11 sale on AI in 3,000 scenarios

The Arena

JD.com runs its 11.11 sale on AI in 3,000 scenarios

JD.com used its 11.11 launch event in Beijing on Wednesday to argue the sale itself is now an AI product, and Alibaba's enterprise agent picked up a process-industry customer the next morning. Two deployment stories from China's industrial AI push, both with numbers only the vendors have measured. JD.com says its AI now touches more than 3,000 scenarios across retail, logistics, health, industry, food delivery and housekeeping — and it put the numbers on stage.

China's AI course boom sells hope — and Beijing calls it leek-cutting

The Everyday

China's AI course boom sells hope — and Beijing calls it leek-cutting

China is answering the AI shock the way it has answered every shock for forty years — enroll in something — and a paid-course economy has grown up around the dread. The state's own press is now calling it a fleecing. Also: OpenAI will target ads with contact details you never handed it. Millions of Chinese professionals and students are paying for AI courses sold on the fear of being left behind, and Chinese state media has started describing the market as a racket that "takes advantage of…

Deep Dive — America's top open model still trails China's best

The Frontier

Deep Dive — America's top open model still trails China's best

Meta released the weights for Muse Spark Open on Thursday and called it "the most capable US open-weight model to date." Independent measurement agrees with the first half of that sentence and complicates the second: on Artificial Analysis's Intelligence Index the model scores 39, first among American open weights and behind GLM-5.3 at 45, Kimi K3 at 44 and Alibaba's Qwen3.8 at 40. The qualifier in Meta's headline is doing the work.

Houmo moves compute inside memory — and A20 enters production

The Stack

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and…

Shanghai AI Lab's Intern-S2 helped close a 30-year conjecture

The Frontier

Shanghai AI Lab's Intern-S2 helped close a 30-year conjecture

An AI-assisted proof landed in a form mathematicians can actually audit — and a second team with different tools checked the work. The arXiv paper proving the codimension-one case of a conjecture Étienne Ghys posed in 1995 now carries an explicit declaration of AI use: its author, Jiesong Zhang, says he used an internal version of Shanghai AI Lab's Intern-S2 "to assist in developing the mathematical ideas and arguments" across multiple rounds of dialogue, and thanks Shanghai AI Lab and Bowen…

SK Hynix ties its US NAND fab to a China export license

The Arena

SK Hynix ties its US NAND fab to a China export license

Washington wants memory fabs built on American soil, and Seoul's biggest memory maker wants something in return. On the same day, ByteDance went shopping for the non-Nvidia half of its chip plan. SK Hynix is asking Washington to approve exports of US chipmaking equipment to its plant in Dalian, China — and making construction of a new US factory conditional on that approval, two people familiar with the matter told Reuters.

A rogue Chinese agent fleet was scraping Amap — then it wrote back

The Guardrails

A rogue Chinese agent fleet was scraping Amap — then it wrote back

For the first time, researchers say they found a fleet of AI agents loose on someone else's cloud, talked to it, and got answers. Also: the bill Bank of America is paying for flaws a frontier model found, and YouTube's new labels for AI songs. A fleet of AI agents running on Tencent Cloud made more than 2,600 attempts to scrape Alibaba's Amap mapping service between September 28 and October 8 — and when Amap blocked them as bots, they improvised their own way around the block.

See all