The letter

Once a week: one article, five AI news items read and commented. Every letter sent, newest first.

Letter no. 5 November 7, 2026 The November 5 AI shock is not what you think Not regulation, not a crash: memory and debt. And Herbert on Apple's chips, against MLX.
  • One level down: #004 the same compiler on Apple's chips
  • On November 5, Wall Street dumped AI: what is measured, what is feared
  • OpenAI produced 948 root exploits on Linux for 1.2 million dollars
  • OpenAI measures that Astra works less well as soon as the task is about safety
  • METR measures that Anthropic is advancing twice as fast, and the evidence is redacted
  • Qwen 4 announces itself, Hydrogen-1.0 is there to download: the week of Chinese models
 Letter no. 4 October 30, 2026 Qwen 4 is expected in early November Six reads this week. And Herbert against TensorRT-LLM on CUDA.
  • One level down: #003 what CUDA has that Vulkan lacks
  • Qwen 4 is expected in early November, its architecture has been public since August
  • Qwen 27B models descend from the 9B: repeated layers, looped Transformers, and a JetBrains blend
  • Google Research gives a sleep phase to language models
  • A simple RAG beats most agent memories
  • OpenAI fired three safety researchers after Signal conversations
  • Hinoki promises an offensive model with no refusals, Cohere catches Qwen aiming at Beijing from a sandbox
 Letter no. 3 October 23, 2026 Quad9 leaves France on November 24 Four more reads, and Herbert against vLLM and llama.cpp on five GPUs.
  • One level down: #002 the same compiler on a GPU
  • DeepSeek opens DeepInfer but keeps distributed orchestration under lock
  • PHP, Node.js, Java, pip, and OpenCode: the week everything needed an update
  • Quad9 leaves France because it cannot geoblock its DNS resolvers
  • Meta opens the weights of Muse Spark: 816 billion parameters and two Mac Studios to run it
  • Anthropic puts 5% of its shares in a public benefit fund, months before going public
 Letter no. 2 October 17, 2026 The Linux kernel, coded with Claude Code Four more reads too, and my engine against llama.cpp, vLLM and OpenVINO.
  • One level down: #001 a compiler for an LLM engine
  • Linus Torvalds signs Linux kernel commits with Claude Code
  • Claude Haiku 4.5 submitted a fictitious report to the Philadelphia police
  • Karpathy: five thousand insiders see an AI that six billion ignore
  • HBM strangles standard DRAM and drives down smartphone sales
  • Three states restart from the same silicon without asking anyone for permission
 Letter no. 1 October 10, 2026 Why an AI believes 9.11 > 9.9 It was reading a Bible verse. Plus 4 other reads, and my engine that writes its own code.
  • One level down: #000 an inference engine that writes its own code
  • Google and the transformer: eight authors gone, $2.7 billion for a return
  • 9.11 vs 9.9: the model that read a Bible verse
  • Irregular: three labs, one door left open to the internet
  • DeepSeek V4.1 Flash: an open model four times cheaper and not for a laptop
  • OpenAI agents used an old German wiki as a messaging system