Running local LLMs on your Mac: what fits, what's free, and what's overkill

Modern Macs can efficiently run language models locally thanks to unified memory architecture, with model size recommendations ranging from 1-3B parameters on any Apple Silicon Mac to 70B+ on machines with 64GB RAM. Free tools like Ollama and LM Studio make this accessible to most users, though smaller, well-tuned models often outperform larger generic ones for specific tasks. The author notes that while local models don't match frontier cloud models in capability, they offer practical benefits for everyday computing without requiring new hardware purchases.