Build an LLM or AI agent stack

Builder

Step 1 of 4 · Local LLM

Your machine

No account needed. Tell us what you'll run it on, and we check fit against the local-model roster.

Memory

Usable for models≈ 18 GB

Running estimate

Local LLM

Apple Mac mini (M4) · 24 GB: hardware you own, no rental	$0.00
Gemma 3 27B: local model, no token bill	$0.00
Estimated total	$0.00/mo

typical $0.00–$0.00

Notes for this configuration

A local machine isn't always-on unless it stays awake and connected.
The only genuine cost cut here: hardware you already own, gated by fit.
Fit is tight (~18 GB usable vs 18 GB needed). Expect slower output.
Account verification is strict and free capacity varies by region; provisioning can take several attempts.
The free Personal plan is limited to non-commercial use (individuals, up to 6 users); team and business use requires a paid plan (Standard ~$8/user/mo, Premium ~$18/user/mo). This is not a universal $0 entitlement.
Tailscale Serve keeps a service private to your tailnet; Tailscale Funnel exposes it to the public internet. They are distinct features and should not be conflated.
Keep access private. Reach your assistant through a secure tunnel, never the open internet.

Loading the latest dated prices…