The Frontier - SiliconNoon (Page 16)

The Frontier

Shanghai AI Lab chief: science is the next programming

One of China's biggest AI labs says the frontier after code is the wet lab — and it shipped a robot-running world model to make the case. Wenbow Zhou, director and chief scientist of the Shanghai AI Laboratory, used the main forum of the 2026 Pujiang Innovation Forum on September 12 to argue that "scientific research is the next programming." His reasoning: programming is AI's biggest real-world productivity win so far, and research has the same shape — problem-oriented, cognition-driven — exce

The Frontier

Fable 5.1 leads Real-SWE, and still fails 6 of 10 tasks

Two benchmark results landed the same day, and both point at the same gap: agents score well on code they have seen, and badly on code they have not. The best coding agent in the world resolved 38.8% of real enterprise engineering tasks — and no agent solved every task in the set. Specific Labs, a Y Combinator-backed evaluation shop, released Real-SWE this week, a benchmark whose defining property is that its code has never touched the public internet. Every task was lifted verbatim or adapted

The Frontier

A 33B model from Singapore just matched DeepSeek V4 Pro — on free tokens

Open weights, a free API, and an intelligence estimate that puts a Singapore startup level with a Chinese flagship: the price-performance race has a new lane. Plus: agents that hustle humans to stay switched on. Agnes-3.0-Flash shipped weights under Apache 2.0, and its estimated score ties DeepSeek V4 Pro. The 33B multimodal model from Agnes AI — the commercial arm of Singapore's Sapiens AI, founded by Bruce Yang — landed on Hugging Face with a 262,144-token context window, text/image/video inp

The Frontier

NASA and IBM open-source a foundation model of the Moon

AI for science keeps climbing the priority ladder — today it reaches another world, while China's compute buildout shrinks to the size of a shipping container. NASA and IBM released an open-source foundation model trained on decades of lunar data, and the agency says it already beats purpose-built models at finding ice. The NASA-IBM Lunar Foundation Model harmonizes observations from the Lunar Reconnaissance Orbiter, GRAIL and Lunar Prospector into one pre-trained representation, so planetary s

The Frontier

Motus2: a robot world model that grades its own actions

Shengshu Technology, the Chinese world-model lab, shipped the rare robotics release where the loop actually closed: one model that acts, imagines, and scores itself — then learns from the grade. Shengshu released Motus2, a "self-evolving" general world model for dexterous manipulation that puts three capabilities — acting, predicting, and evaluating — inside one shared-parameter video-action model, according to the project page and the paper on arXiv. A policy proposes candidate action chunks,

The Frontier

Google's TimesFM-3 forecasts every variable at once

A quiet-but-real model release out of Google Research, and a state-backed number for just how much computation the agent era is expected to eat. Google's TimesFM-3 became the first model in the family that can forecast many variables at once — and it fills the entire future in a single pass. The 330-million-parameter transformer chops each series into patches of 32 points, normalizes them onto one scale, then alternates two attention directions: along time within a series (past values only, so

The Frontier

Deep Dive — Tier 4 fell in 14 months. Nothing can refill it

The hardest closed-form math benchmark in the industry is over, and the interesting fact is not the winning score. It is that every candidate replacement is structurally worse at the one job a benchmark has to do: letting someone outside a lab check whether the lab is telling the truth. As this morning's brief reported, Epoch AI now treats FrontierMath Tier 4 as saturated. GPT-6 Astra cleared the final unsolved item — a research-level question written by combinatorialist Jay Pantone — taking th

The Frontier

GPT-6 Astra solves the last FrontierMath Tier 4 problem

A research-grade math benchmark built to stay ahead of models just ran out of road, Apple's dedicated neural chip gets its autopsy, and Oracle puts a price on the AI jobs it is cutting. Epoch AI's FrontierMath Tier 4 is saturated. GPT-6 Astra solved the final unsolved problem — a research-level question authored by combinatorialist Jay Pantone — and Epoch now considers the tier closed out at 97.6%. When Tier 4 launched in mid-2025, the best model cleared just 5% of it; the runners-up are GPT-5.

The Frontier

Kimi K2.8 Preview puts near-K3 coding in every membership tier

Moonshot spent the week arguing about revenue and the weekend shipping a model that undercuts its own flagship. Meanwhile the person who built Redis quietly made DeepSeek's newest MoE run on a desktop Mac. Moonshot AI put Kimi K2.8 Preview live across Kimi Code and Kimi Work on September 11, claiming overall performance close to flagship K3 with further gains in coding and agentic work, and a 1M-token context window open to every membership tier. The pricing gap is the point: K3 needs the ¥99-p

The Frontier

An open recipe for IMO gold: Nemotron scores 30/42 at IMO 2026

A quiet but consequential morning for machine mathematics: Nvidia has published the full recipe behind an open model reaching the International Mathematical Olympiad gold-medal threshold, and two new papers push at the boundary between informal reasoning and machine-checked proof. Nvidia's Nemotron scored 30 out of 42 points at IMO 2026 — officially graded, above the gold-medal cutoff of 29 — and the company has released everything: the two post-trained model checkpoints, the training data, the

The Frontier

Apple's SimpleDesign ditches the latent space for protein co-design

Apple's ML research group has had a quietly productive week, and today it landed three papers at once: a protein-design model trained straight in the raw data space, a benchmark that grades video captions with quizzes instead of reference text, and a sign-language translator that finally pays attention to what was said two sentences ago. None is a consumer launch, but together they sketch where Apple's research muscle is pointed — biology, evaluation, accessibility. Apple's SimpleDesign generat

The Frontier

Loop a transformer twice, save 18% of training compute

Two stories from the research world today: an architecture paper that puts hard numbers on one of this year's more interesting ideas in model design, and a peer-review system that is quietly rewriting the economics of who gets to submit. Re-running a transformer's middle layers twice can cut training compute by up to 18% — if you pay for the extra passes honestly. That's the finding of SMELT (Sparse MoE Transformer, middle layers Loop Twice), a paper from teams at Tsinghua University, ByteDance

See all