New models, papers, and benchmarks
The Frontier
Brain-to-text research usually leans on MEG machines, fMRI tubes, or electrodes surgically placed inside the skull. A paper posted to arXiv this week makes the case that a dry-electrode headset sitting on the scalp — the kind of hardware that costs hundreds of dollars, not millions — can recover
The Frontier
Two useful signals surfaced on HN today: Stanford's updated verdict on what AI is actually doing to jobs, and a prediction-market dashboard trying to put dates on the next frontier model releases. One deflates the panic; the other fills a vacuum the labs created. Erik Brynjolfsson says
The Frontier
A new theoretical paper argues the real risk of AI in science isn't that models make mistakes — it's that they make researchers' time too valuable to spend polishing anything. Even a perfect LLM, the authors say, could push scientists toward shallow output. A team from
The Frontier
The 2.8-trillion-parameter open model that most GPU fleets can't hold turned out to be hostable on a single rack — and the community is now arguing over what that really costs. A developer posted a full end-to-end deployment of Kimi K3 on eight NVIDIA
The Frontier
A learned kernel is the low-level math software that runs on a GPU and was designed — or tuned by trial and error — by an AI program, rather than hand-written by a human engineer. When a large language model does its math, it doesn't just "think&
The Frontier
Small models are quietly winning the jobs nobody brags about. Liquid AI says its tiniest LFM2.5 release yet — a 230-million-parameter model — outperforms rivals four times its size on data-extraction work, and it is small enough to run almost anywhere. Liquid AI's new LFM2.5-
The Frontier
The mystery model that appeared on OpenRouter this week with no name attached is still free, still anonymous, and now the center of a very public argument about whether it's any good. Early benchmark results are pointing in opposite directions — and the fight over Ox Alpha's
The Frontier
Mechanistic interpretability rarely gets a result this clean: switch off a single head, and a chess AI loses Morphy's most famous move. Silencing one of a chess transformer's 128 attention heads makes the model stop finding the Opera Game queen sacrifice, according to a project post
The Frontier
Seven days after Alibaba dropped a frontier-grade coder onto consumer GPUs, the community has had time to do what it does best: benchmark it, compress it, wire it into every inference stack available, and argue about what it means. One week after Qwen 3.8 27B landed as a
The Frontier
Two releases landed tonight that say a lot about where AI is heading: a benchmark that finally measures whether models can do research rather than talk about it, and DeepSeek making its API cheaper exactly when nobody is watching. Prime Intellect's NanoGPT Speedrun Frontier is the first public,
The Frontier
A London lab founded by Google DeepMind alumni says its new agent, Faraday, outperformed Claude Opus 4.8 and GPT-5.5 at independently reproducing the findings of published scientific papers — and it did it on a 27B-parameter Qwen 3.6 model, a fraction of the frontier systems'
The Frontier
Richard Sutton, the reinforcement-learning pioneer who wrote "The Bitter Lesson," is one of the few people in AI whose contrarianism is worth taking seriously. But his most cited claim this week isn't his strongest. He is wrong to call synthetic data the industry's