- 411The Hackintosh Reborn (naut.ca)
- 489Discord is ditching Electron and its new app uses less RAM (theverge.com)
- 30Flawless UI-UX Skill 1.0.0 (github.com)
- 97The Stacks Project, Formalized in Lean (axiommath.ai)
- 17Lighting the Skies (computer.rip)
- 23Half of Show HN posts are a bare link. Ones with 500 characters got 2x replies (ownfeed.dev)
- 558I wish tech companies knew more about me (rakhim.exotext.com)
- 30Show HN: Agents infer Lego mechanisms from sources, then animate them (github.com)
- 636Kimi K3.1 Pro model weights are out (huggingface.co)
- 5He Was a Suspect in the Yogurt Shop Murders for Thirty Years. He Can't Forget (texasmonthly.com)
- 751Search for everyone (kagi.com)
- 287Inkling 2 Small (thinkingmachines.ai)
- 341GPT-6 Astra covertly underperforms on safety research (alignment.openai.com)
- 147Being a kid making games in the early 2000s (jaytholen.net)
- 151A wave of plastic-free coffee pods and machines (cnn.com)
- 578Motorola Signature 27 with GrapheneOS support (motorolanews.com)
- 105Microsoft is winding down Oracle deal to rent AI cloud capacity (reuters.com)
- 103Building a Multi-Agent System from Scratch: Architecture Patterns (kadhar.dev)
- 13Project Sync: The journey to sample-perfect audio/video sync with WebCodecs (mediabunny.dev)
- 134University libraries are being swarmed by AI bots, putting public access at risk (cbc.ca)
- 17Why Amstrad chose the 3" drives (cpcwiki.eu)
- 28AI coding agent setups from 663 Hacker News comments (theainewsreport.com)
- 62Perplexity launches AI agents for work (perplexity.ai)
- 170'It's just vape, nail, charity shops': Can anything save UK high street? (theguardian.com)
- 728Tell HN: Gemini trains on chat history, even for subscribers (personal accounts) ()
- 262Upload Your RAM (uploadyourram.org)
- 112We can do better than line-buffered output (nullprogram.com)
- 269September in Servo: <progress>, rounded rectangles, and more (servo.org)
- 281Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights) (linum.ai)
- 615Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux (mindstudio.ai)