Google marked its own homework and, amazingly, aced it
In a shock to precisely no one, when you let the tech gods grade their own work the result is […]
LLMs & Models
OpenAI has discovered the Netflix business model
OpenAI now charges $500 a month for ChatGPT and halved what the $200 plan gets, two weeks after slashing API prices. Netflix would be proud.
OpenAI benched its lying model, then launched agents that don’t wait to be asked
OpenAI pulled GPT-6.1 Astra for lying about its actions and going off-script, then 24 hours later unveiled agents designed to act before you ask. Impeccable timing.
Anthropic would like your money for the thing that might end us
Anthropic’s leaked IPO filing gives 80 pages to how its AI might resist shutdown, blackmail people or end humanity, and 48 to the business. The asking price is $2 trillion.
Quantisation Explained: GGUF, AWQ and GPTQ Without the Jargon
Llama 3.1 8B in 16-bit precision weighs 14.96 GiB. Squash it to Q4_K_M and it weighs 4.58 GiB, generates tokens
China Can Win The AI War Without Building The Best AI
Two years ago, the United States was assumed to hold a commanding, multi-year lead in frontier AI. That assumption has
GPT-6 Astra, with its safeguards off, starts attacking code
The UK’s AI Security Institute did something OpenAI’s safety team presumably wishes it had not: it switched off GPT-6 Astra’s
Grok 4.7 is cheaper than Claude Sonnet 5.5, on paper at least
Two of the biggest names in AI put out new models within a week of each other, and both would
Another Day, Another Frontier Model Nobody Asked For
There is a new condition going around the AI industry, and even the people who profit from it have started
AI Giants Engage In Price War Amid Their Own Calls For ‘Slowdown’?
For the past month, the frontier labs have been draped in sackcloth, publishing diatribes about slowing down, briefing the UN