✓It Still Works
  • Home
  • Search
  • Discover
  • Catalogue
  • Categories
  • What changed
  • Developer repos
  • How we check
  • Coding and upskilling
  • Developer utilities
  • Draw and diagram
  • Everyday tools
  • Games and play
  • Images and photos
  • Music and audio
  • Save and convert media
  • Testin category
  • Writing and publishing
  • Cloudflare (4)
  • Vercel (3)
  • GitHub Pages (2)
  • Google Cloud (2)
  • nginx (self-hosted) (2)
  • Google (1)
  • Netlify (1)
712 checks loggedEvery listing, every sweepSign in
Theme
✓It Still Works
Sign in

Every listing is checked on a schedule. No numbers we cannot measure.

How we checkGraveyardSubmit something
  • Discover
  • Catalogue
  • Changes
  • Repos

I understood

run a frontier-sized model on hardware you already own

Not what you meant?

  • audit model and tool use on your own infrastructure
  • run a personal AI agent on your own machine rather than someone else's
  • run several coding agents at once without them colliding
  • give each agent its own branch through a git worktree

Tools(5)

run a frontier-sized model on hardware you already own · run agents with your real dev environment rather than a hosted sandbox · use a cheaper model behind a coding CLI you already know · run coding agents in the browser on your own machine

5 listings shown. Every result was live at its last check.

  1. Run a model far larger than your machine by streaming its weights off disk

    Why this one

    • run a frontier-sized model on hardware you already own
    • try a large model without renting a GPU box
    • read how streamed expert weights work in plain C

    On the record

    • Free
    • No signup

    Last checked 1 day ago

    Open↗
  2. Run your coding agents on your own machine and drive them from your phone

    Why this one

    • run agents with your real dev environment rather than a hosted sandbox
    • check on a coding agent from your phone while away from the desk
    • kick off a job before you leave and read the result later

    On the record

    • Free
    • No signup

    Last checked 2 days ago

    Open↗
  3. Run Codex CLI and Claude Code against whichever model you want

    Why this one

    • use a cheaper model behind a coding CLI you already know
    • route different kinds of request to different models automatically
    • keep using Codex when the model you want is somewhere else

    On the record

    • Free
    • Your own API key

    Last checked 2 days ago

    Open↗
  4. Run Claude Code, Codex and 11 more from phone, browser or desktop

    Why this one

    • run coding agents in the browser on your own machine
    • drive Claude Code from a phone

    On the record

    • Free
    • Needs an account
    • Your own API key

    Last checked 2 days ago

    Open↗
  5. Run several coding agents at once, each in its own git worktree

    Why this one

    • run several coding agents in parallel without them treading on each other
    • review what an agent changed before it reaches your branch

    On the record

    • Free tier
    • Needs an account
    • Your own API key

    Last checked 2 days ago

    Open↗