Ultrathin business card runs a fluid simulation

I want everything local – Building my offline AI workspace

Jim Lovell, Apollo 13 commander, has died

Sam said yesterday that chatgpt handles ~700M weekly users. Meanwhile, I can&#x27;t even run a single GPT-4-class model locally without insane VRAM or painfully slow speeds.<p>Sure, they have huge GPU clusters, but there must be more going on - model optimizations, sharding, custom hardware, clever load balancing, etc.<p>What engineering tricks make this possible at such massive scale while keeping latency low?<p>Curious to hear insights from people who&#x27;ve built large-scale ML systems.

Ask HN: How can ChatGPT serve 700M users when I can't run one GPT-4 locally?

The surprise deprecation of GPT-4o for ChatGPT consumers

Getting good results from Claude Code

Tor: How a military project became a lifeline for privacy

How we replaced Elasticsearch and MongoDB with Rust and RocksDB

Window Activation

Astronomy Photographer of the Year 2025 shortlist

How attention sinks keep language models stable

Apple's history is hiding in a Mac font

Build durable workflows with Postgres

Efrit: A native elisp coding agent running in Emacs

Fire hazard of WHY2025 badge due to 18650 Li-Ion cells

Telefon Hírmondó

<a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=TaYVvXbOs8c" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=TaYVvXbOs8c</a><p><a href="https:&#x2F;&#x2F;github.com&#x2F;langchain-ai&#x2F;open-swe" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;langchain-ai&#x2F;open-swe</a>

Open SWE: An open-source asynchronous coding agent

Little-known leguminous plant can increase beef production by 60% (2022)

I bought a £16 smartwatch just because it used USB-C

Hacking Diffusion into Qwen3 for the Arc Challenge