4 ms·

Mac Minis are perfect for locally running demanding models because they can effectively use ordinary RAM as VRAM.

• hjoutfbkfd 9mo ago
  but people dont use OpenClaw with local models
  • paunchy 9mo ago
    They definitely do. A common configuration is running a supervisor model in the cloud and a much smaller model locally to churn on long running tasks. This frees Openclaw up to lavishly iterate on tool building without running through too many tokens.
    • garciasn 9mo ago
      Unless you're running a large local model in 192GB+ this just won't be ideal, based on real-world experience.