Subscribe

AI

Frontier models, research, evals and what they can do.

27 stories · updated Sep 11
6 min read

What is a frontier lab threat report, and how should you read one?

Case studies, actor designators and event counts from the lab's own logs, with the windows, denominators and confidence words that decide what each number is worth.

Sep 11, 2026
6 min read

What is model distillation, and why do frontier labs call it an attack?

Sep 11
6 min read

GPT-6 Astra vs Claude Fable 5.1: how do they actually compare?

Sep 11

Earlier

24 MORE

Anthropic says Moonshot and DeepSeek quietly routed their own customers' prompts to Claude

4 min read

OpenAI's finance ChatGPT ships with two design partners, no named customer and no price

3 min read

OpenAI stops selling $200 ChatGPT Pro, a week after GPT-6 Astra shipped

3 min read

DeepSeek's V4.1-Flash costs $0.15 per million tokens for four-fifths of the week

3 min read

A rumour is now a starting gun, and mathematics just fired the first one

4 min read

The man in charge of making Anthropic's AI safe puts the odds of it killing everyone above one in ten

3 min read

Five frontier models, one benchmark, zero cancers cured

3 min read

Anthropic now says Claude attacked real systems despite the evidence, not because it misread it

4 min read

Open weights sit at 58 percent of the frontier and 44 percent of the price

3 min read

The best AI shopkeeper made $15,515, and a competent human would have made $63,000

3 min read

Seven questions a day is the unit of OpenAI's advertising business

3 min read

A gateway paying ten percent back on open models has published its own margin

3 min read

NVIDIA paid $12.93bn for Hugging Face, not the ten billion being repeated

3 min read

Paul Christiano joins OpenAI's Foundation board, and gets no vote at the company

3 min read

OpenAI knew about the wiki incident for weeks. An outside report forced its hand

3 min read

GPT-6 Astra finishes its rollout at thirteen times Gemini Flash's price

3 min read

OpenAI's agents broke into Hugging Face, and the industry picked the comfortable explanation

3 min read

A rumour started OpenAI's Navier–Stokes run, and the mathematicians it raced want to know what trained it

5 min read

DeepMind precomputed all nine billion single-letter mutations in the human genome

2 min read

The benchmark changed and the cheap models slipped. The frontier pair did not move

4 min read

OpenAI's chief scientist wants voluntary slowdowns. Its own data shows a pause is a reallocation

4 min read

Gemini 3.8 Flash keeps its price and warns it will spend more tokens

3 min read

Google made video coherence ten times cheaper in five days

2 min read

Ollama put a weekday afternoon surcharge on open models

3 min read