The Frontier - SiliconNoon (Page 8)

The Frontier

Deep Dive — Inside the sandbox factory that trains DeepSeek's agents

This morning's brief covered the headlines; this is the machinery underneath. On September 19, DeepSeek published a systems paper with an unglamorous title — "DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale" — and an unusually revealing body. DSec is the platform that served every sandbox workload in the company's RL training and evaluation from DeepSeek V3.2 through V4.1. The byline carries well over a hundred names and ends with founder Liang W

The Frontier

DeepSeek shows the machine that trains its agents — and it runs 3 million sandboxes a day

While Western labs debate whether their agents are safe in theory, DeepSeek just published the factory floor. A new paper, "DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale" (arXiv:2609.22978), lays out the production system behind the lab's agentic reinforcement learning — and founder Liang Wenfeng is the last-listed author of the 162-author submission. The scale is the story: a single production unit spans roughly 160 nodes serving about 3 milli

The Frontier

Stanford ran a virtual biotech of 37,000 agents

A Stanford team ran 37,075 AI agents as a "virtual biotech" for a week, and the therapeutic strategy it proposed for lung cancer is one Merck and Daiichi Sankyo independently took to an FDA breakthrough designation last year — but no molecule came out of it, and nothing was tested. The paper, published in Science on 17 September by Harrison Zhang, Peter Eckmann, Jiacheng Miao, Andrew Mahon and senior author James Zou, points an organisation of more than 37,000 agents at 55,984 annotated clinical

The Frontier

Ex-Huawei Pangu CTO raises hundreds of millions for a physics model

Two of Huawei's best-known large-model people have a new company, and it isn't Huawei's. Shenzhen's XIRRA has raised several hundred million yuan to build a foundation model for the physical world — and the framing around it needs one correction. Li Yin, the former CTO of Huawei Cloud's Pangu large-model team, left to found XIRRA (Shenzhen Xirrang Kaiwu Technology), and the company has now closed seed and angel rounds worth several hundred million yuan in total. Dunhong Assets led, with Huakong

The Frontier

Lambert's Hill briefing: Chinese open weights out-download US 2-to-1

Nathan Lambert's Hill briefing: Chinese open weights out-download US 2-to-1 Three stories about who sets the terms: the open-weight balance sheet Congress got briefed on privately, the banks trying to write the rules for shopping agents before a regulator does, and the clinics getting frontier medical AI for free. Chinese open-weight models now hold twice America's cumulative downloads and more than 80% of open-model usage on OpenRouter, according to the remarks Nathan Lambert prepared for a c

The Frontier

OpenAI halves GPT-6 Sol and Luna prices hours after Anthropic's cut

Two frontier labs spent Tuesday afternoon arguing about the same thing — not who is smartest, but who costs least per task. OpenAI cut API prices for GPT-6 Sol and Luna in half on Tuesday, hours after Anthropic cut Opus 5.5 pricing by 40%. Sol drops to $2 per million input tokens and $10 per million output, from $4 and $20; Luna goes to $0.10 and $0.50 from $0.20 and $1.20, measured against the promotional pricing of the GPT-5.6 versions they replace. OpenAI attributes the cut to better prompt

The Frontier

Anthropic's Opus 5.5 matches Fable 5.1 for 40% less than Opus 5

Anthropic's first release since Dario Amodei called for pacing the frontier is not a capability jump. It is a cost cut — with a guardrail bolted on. Anthropic released Claude Opus 5.5 on Tuesday, saying it performs at the level of the larger Fable 5.1 on most work and costs 40% less to run than Opus 5. Output tokens drop to $20 per million from $25, input to $4 from $5, and the model uses fewer tokens per task on top of the cheaper rate — which is where the 40% figure comes from. Five-hour usag

The Frontier

Inspur says one box now runs 2.8T Kimi K3 on 128 chips

The largest open-weight model in existence has been given a machine that fits in one rack — and there is no Nvidia silicon inside it. What Inspur actually put on stage At the AI Computing Conference in Beijing on September 21, Inspur Information launched the Yuanbrain SD200 Ultra supernode alongside a second product, the Yuanbrain HC2000 multi-compute rack. The claim for the SD200 Ultra is compact: it runs Moonshot's Kimi K3 — 2.8 trillion total parameters, 104 billion active per token — on a

The Frontier

Lab-grown neurons gave a video model a 156,000-parameter edge

Two stories today: a Baltimore startup turned recordings of living brain cells into a shippable AWS product, and Chinese labs got a seat at the UN Security Council's AI table. The Biological Computing Company is now selling its neuron-derived optimizer through AWS — and the neurons are the least commercial part of it. TBC grows cortical cells on a silicon chip with 4,096 electrodes, feeds them image data as electrical pulses, records how the excitation spreads across the culture and fades, then

The Frontier

Kimi Delta Attention gains 2D rotations from a signed gate

A two-number change to one of the open field's most widely used attention modules buys a provable class of expressivity at no extra cost — and the authors are unusually clear about what it does not buy. Allowing Kimi Delta Attention's channel-wise gate to go negative — into [−1, 1] — and its delta-rule coefficient up to 2 lets the module realise 2D rotations, giving it the state-tracking expressivity of DeltaProduct-2 while its transitions stay diagonal-plus-rank-one and non-expansive. The pape

The Frontier

Orbital inference starts selling: Yusuan Xinghe ships in October

In-orbit AI stopped being a demonstration this week. One company put a launch date and a price list on it, and a second put the hardware on orbit to make it work. A Guangzhou startup says its first compute satellites launch in October, and customers have already paid in advance for the models they will run. Yusuan Xinghe, incubated by listed smart-transport firm PCI Tech (Jiadu Technology 600728), used the second Guangdong AI Application Matchmaking Conference in Guangzhou on Tuesday to unveil

The Frontier

Tencent ships Hy Image3.5 at $0.024 an image — parity claim untested

Tencent's image line has been on a two-month clock since it rebuilt the stack in February. The latest tick is a preview model whose rate card does more work than its benchmark does. Tencent released Hy Image3.5 preview on Tuesday, pricing a 2K image at 0.15 yuan on its domestic cloud API and $0.024 on the international endpoint — about a quarter of what the model it claims to match charges. The model does text-to-image and image-to-image, takes up to five reference images per call for identity

See all