2026-09-22 xai
Jared Palmer's Kev, tiny Qwen3.5 decision models, tops HN (367 upvotes, 164 comments); xAI ships Grok 4.7 claiming 2x speed at half price while third-party tests rank its output speed near the bottom (423/343); the Snowden archive has had zero new documents in seven years, with ~99% never published (663/477); ZuckOff spots Meta smart glasses before they record you (587); npm package mathmain posed as a math library to ship an encrypted implant; the M5 Ultra Mac Studio tested as a local-agent machine with up to 512GB unified memory at 1.2TB/s; Cory Doctorow's 'Claude Delusion' draws nearly twice the comments of upvotes; Apple's own docs explain how to turn off Apple Intelligence.
Read analysis 2026-09-21 openai
A parody site topped HN by asking AI agents to upload their own weights (587 upvotes, 242 comments); a researcher shows ChatGPT tracking users across 936 advertiser sites via a measurement cookie (424 upvotes, 224 comments); Alibaba's Qwen Image 2.1 packs 7B parameters with native 2K and transparency under a research-only license; a cryptographer factored RSA-896 with Claude as collaborator; Microsoft's agents ported the Copilot runtime to Rust for $120K, a 15.9x throughput gain at 1/11 the memory; Samsung plans to double HBM4 output; StepFun's Step 5 Preview posts 600B params at $1/$2.70 per million tokens; Sam Altman heads to the UN Security Council; and self-hosted inference orchestrators compared.
Read analysis 2026-06-16 ollama
Vicki Boykis says local models are good now. A 1,245-point Ask HN thread splits into two camps. Boosters measure whether local open-weight models handle daily coding. Skeptics measure whether they match cloud frontier models on hard tasks. The turning point is not that models suddenly got smart, it is that open weights crossed a usable line and local agent tooling redefined good enough. The builder question: not can it work, but how far apart are success rate, latency, and cost on your actual tasks, and is the gap worth trading privacy and control for.
Read analysis 2026-06-11 google
Google open-sourced the first mainstream text diffusion model. The real story isn't 'fast'. It's that the local decode bottleneck moves from memory bandwidth to compute, with bidirectional attention generating 256 tokens at once. The cost: quality, experimental status, and the 26B MoE trade-offs.
Read analysis 2026-06-11 google
Gemma 4 12B feeds vision and audio straight into the language backbone, dropping dedicated encoders. That's an architecture bet, not just another on-device model.
Read analysis 2026-06-11 google
Google shipped quantization-aware training weights for Gemma 4, squeezing E2B down to 1GB so it runs on phones and consumer GPUs. The turn that matters isn't 'it fits now'. It's that the hard problem moved to power draw, the privacy boundary, and exactly how much quality you lose.
Read analysis 2026-06-11 hcompany
H Company ships its first computer-use model you can run locally. It does not chase the top of the leaderboard; it tackles the problem cloud setups cannot escape: every step ships your screen out.
Read analysis