Mac mini, MacBook Pro split local LLM duties

This is a workflow question more than a hardware question: if you live in cafés, the laptop-first setup removes friction, but if you can tolerate a VPN and occasional latency, the always-on home box is the cleaner LLM appliance.

• –Tailscale’s docs explicitly frame tailnet access as a way to reach a private server from anywhere over encrypted connections: https://tailscale.com/docs/concepts/local-team-server
• –The home-machine setup wins for 24/7 inference, longer jobs, and keeping sensitive data off a mobile device
• –The MacBook Pro wins when you need an always-available interactive machine and do not want to depend on network quality
• –The real pain point is not connectivity setup, it is latency, session continuity, and what happens when you lose the connection mid-task
• –For data science and local LLMs, the hybrid model is usually strongest: light laptop for mobility, heavier home rig for batch work and local model serving