The Frontier - SiliconNoon (Page 24)

The Frontier

Deep Dive — Marin's 535B run puts open model training on livestream

A 535-billion-parameter model is training right now with its code, data mixture, loss curves, and even its failures posted in public as they happen. The Marin project started the run this week and wants the world to watch the experiment unfold rather than read about it in a retrospective paper three months from now. It is the starkest test yet of a question the open-weights community has never fully answered: can transparency at frontier scale be real, or is it just branding? We covered the lau

The Frontier

Marin opens a 535B-model training run in live public view

A Chinese lab just started what it bills as the largest fully-open large-model training run ever — and it wants the world to watch it happen in real time, not read about the results after the fact. Marin has begun training a 535-billion-parameter MoE model with its code, data mixture, and live loss curves published as it goes. Called Marin 535B-A23B, the model activates about 23 billion of its total parameters per token and is being trained on 18.75 trillion tokens across 11 NVIDIA GB200 NVL72

The Frontier

China submitted 60% of IJCAI 2026 papers — then couldn't attend

IJCAI-ECAI 2026 closed in Bremen over the weekend with an awards slate that reads like a map of AI's past and future — and a participation report that exposed a widening gap between who produces AI research and who actually shows up. China submitted more than 4,000 papers to IJCAI-ECAI 2026 — over 60% of the entire conference — yet visa refusals kept a large share of its researchers out of the hall. Program committee chair Diego Calvanese delivered the closing numbers on August 22: Chinese regi

The Frontier

Open-source robot brain DM0.5 takes the crown on RoboDojo

The hardest leaderboard in robotics has a new number one — and this time, anyone can download it. Dexmal's DM0.5, an open-weight vision-language-action model, has taken first place on RoboDojo, according to Dexmal and QbitAI's report on the leaderboard. Its composite score of 24.90 and 19.34 percent average success rate sound modest until you see the field: as of July 2026, models at the top of RoboDojo's simulation board averaged just 8.80 percent success, and the real-robot board 12.8 percent

The Frontier

Claude helps settle a 78-year-old question about the six-sphere

A question that has stumped mathematicians since 1948 — and tripped up some of the field's most decorated names — appears to have finally fallen, with Claude helping a Harvard mathematician build the answer rather than search for it. In a year already full of AI math milestones, this one looks different in kind, not just degree. Also in the inbox: Nvidia's biggest customers start seeing the memory crunch show up as an actual price hike. A Harvard mathematician working with Claude has apparently

The Frontier

Apple cuts video-AI latency 5x by internalizing visual thinking

A new Apple machine-learning paper takes aim at one of the more expensive habits of multimodal models: rendering future frames to "see" where a video is heading. Its move is to push that thinking into training — and drop frame generation at inference entirely. Apple researchers show video models can learn to reason visually during training and then skip frame generation at inference, matching or beating the standard approach while cutting end-to-end latency more than fivefold. The method, descr

The Frontier

ShengShu lays out a five-level roadmap to general world models

ShengShu founder Zhu Jun used a World Robot Conference keynote to put a name and structure on where embodied AI is heading: a five-level roadmap for a general world model he argues must unify generating the world, interacting with it, and acting inside it — not a stack of separate models bolted together. Zhu, the ACM/IEEE/AAAI fellow who built video model Vidu, described the general world model as a closed loop rather than a single generator or robot policy: a system that understands the world,

The Frontier

Kids learn language on a sliver of an LLM's data — we still don't know why

A child picks up its mother tongue after hearing maybe 100 million words. A frontier model needs trillions. MIT Technology Review takes a long look at the divide researchers call the data-efficiency gap, why it's quietly becoming the industry's biggest problem, and what closing it could unlock for everyone who can't afford to hyperscale. A child masters its mother tongue with maybe 100 million words of exposure, while the best language models burn through trillions — and nobody can yet explain

The Frontier

Deep Dive — AI co-reader caught 15 liver cancers radiologists missed

Alibaba's research arm, DAMO Academy, has published a medical-imaging AI model in Nature Medicine that reads the contrast-enhanced CT scans radiologists already order and flags liver cancers the human eye missed. In a two-month deployment inside a routine clinical workflow, the model helped clinicians find 15 malignant tumors — most of them around a centimeter in diameter — that had been overlooked, and changed treatment plans for patients as a result. We should be careful about what this is an

The Frontier

Two unannounced Claude models just surfaced in Anthropic's API — and one may already beat Opus 5

Anthropic's next lineup has leaked the old-fashioned way: developers spotted two never-announced models, claude-marshmallow-eap and claude-melon-eap, sitting in the company's API. Multiple developers confirmed both codenames through API access and Discord, and early testers report that Marshmallow's conversational performance already edges out Opus 5 — Anthropic's current flagship, released barely a month ago. Neither model carries a Fable-class designation, so community consensus points to them

The Frontier

Nikkei's AI scorecard puts Kimi K3 sixth — and Meta poaches another OpenAI researcher

Two signals about where AI's center of gravity sits right now: a Japanese survey says the strongest open model is now beating closed giants on trust — and the talent war for the people who build them keeps escalating. Nikkei's joint survey with Weights & Biases has Kimi K3 climbing to sixth overall, scoring ahead of Google's and xAI's flagship models. The "AI Model Score" comparison, run by Nikkei Digital Governance with the developer-tools company Weights & Biases, ranks Moonshot AI's Kimi K3

The Frontier

Alibaba's Wan3.0 goes live with 30-second clips and doc-to-video

Alibaba's video generator just graduated from beta to a real product — and its two flagship tricks, half-a-minute single-shot clips and slides that film themselves, aim straight at business users rather than filmmakers. Alibaba Cloud officially launched Wan3.0 on Monday, taking its video-generation model out of a public beta that opened in early August — and the production version can generate up to 30 seconds of video in a single pass. Chinese outlets including Sina Finance reported the rollou

See all