About Local First Lab
Privacy-first developer running a fully self-hosted, offline stack: local LLMs on a personal GPU, encrypted storage, and the open-source tools reviewed here in daily use. Every comparison is written from hands-on experience running and maintaining the software, not from vendor copy.
Local-first software and supervised AI, tested on real hardware. That is the whole pitch, and every page on this site is an attempt to show it rather than just say it.
Why pseudonymous?
Practicing what this site preaches. The reviews stand on their specifics (hardware named, costs itemized, failure modes described), not on a byline. Code is public under the same name: github.com/MacadamiaButter.
How this site makes money
Some links are affiliate links (always disclosed on the page). No ads, no trackers, no sponsored placements. Rankings never depend on commissions; see the privacy policy for the full breakdown.
Found something wrong?
Tell me, corrections make the site better: hello@localfirstlab.org
Need this set up for you?
I do self-hosting migrations, privacy stack audits, and private/local AI installs, including for confidentiality-bound firms. Reach out at the address above.
Where this site stands
What your browser just told us
Any site you visit can read a stack of signals the instant the page loads: no cookies, no permission prompt. This page is no exception; the difference is what happens next.
Lab log
2026-07-25: adversarial security pass on lanewatch, proving a hostile local endpoint could repaint the terminal or balloon memory to 1.3 GB, then fixing both and pinning 25 security tests. Before that, 2026-07-17: task-success bench, a 4B and a 36B local model across two conditions (2x2) over 14 scripted browser goals: authoring validity 14/14 everywhere, task success 0-3/14. Wire format, not model scale, drove the dominant failure. Full results →