Run free local LLMs by RAM: Qwen 4B on 8GB, 35B MoE on 32GB, fully offline
alex_verem · x · 2026-11-09
Alex Verem shares a quick guide to replacing paid AI subscriptions with free open-source models running locally:
- 8GB RAM: Qwen3.5 4B for writing, summaries and quick questions
- 16GB RAM: Qwen3.5 9B for everyday chat and drafting
- 32GB RAM: Qwen3.6-35B-A3B for coding and harder reasoning
Grab the 4-bit quantized versions via free apps like Jan or Ollama. Everything runs offline, so nothing you type leaves your machine.
Related event: Pick Local AI Models by RAM to Cut Subscription Costs(2 posts)→
More from Infra
- Atomarine Raises $19M Seed to Build Nuclear-Powered Floating Data Centers — ycombinator · 2026-11-09
- OpenAI details Pando, a Git-compatible backend built for Codex-scale development — rhiever · 2026-11-09
- Data center logic flips: Argentina's Patagonia courts compute with stranded energy — matt_slotnick · 2026-11-09
- Cathie Wood pushes back on data center fears: electricity prices are real, the rest is overblown — CathieDWood · 2026-11-09
- Bipartisan Senate bill could fast-track US AI data centers, vote as early as next week — Polymarket · 2026-11-09
- Larry Page predicts 99.999% of compute will end up in orbit, citing US data center regulation — elonmusk · 2026-11-09