Contribute

See what your PC can do

The network runs on PCs like yours. The local check estimates a hardware tier, then tells you what that tier can host today.

  • Windows 10/11 64-bit today. macOS and Linux are not supported yet.
  • 8 GB RAM minimum; a 4 GB GPU helps, and a 4-core CPU can still contribute.
  • You choose whether the worker runs, the idle-only wait, and active hours. The app enforces those settings locally; CPU, GPU, temperature, and bandwidth ceilings are not exposed in this build.

The check runs locally. Nothing is uploaded until you ask for an invite.

Contribution is invite-only. There is no public signup.

Local check

Check this PC

Four quick steps estimate the tier this PC can host. The check runs in your browser.

We read

  • Browser
  • Graphics adapter or CPU
  • Memory estimate
  • Operating system

The check runs locally. Nothing is uploaded until you ask for an invite.

The browser can't read exact VRAM. The Windows app confirms it after install.

What we detect

Browser and operating system, a graphics adapter string or CPU core count, an approximate memory size, and WebGPU/WebGL support. No hardware strings leave this page.

What a worker PC needs

Minimums are per worker, not per request. The native app re-checks them before install.

Minimum worker PC requirements
Operating system Windows 10/11 64-bit
System RAM 8 GB minimum
Graphics 4 GB VRAM or CPU
Disk space ~5 GB available (LFM3 + app)
Network Broadband; unmetered preferred

Install after an invite

Invited contributors receive the Windows installer and a one-time enrollment token by email. The installer is not publicly linked until it is code-signed.

There is no public installer download, no account signup, and no API key creation on this page.

What the invite covers

  • The model and role the worker will serve.
  • A one-time enrollment token, bound to the install.
  • The limits the worker must respect on the machine.

Do not forward the email or share the token. The operator sends both manually.

Waitlist

Request an invite

Tell us whether you want to call the API, contribute a PC, or both. We review requests manually.

We use this address only to reply about access. Read the privacy notice.

You keep control

The worker runs only while its limits allow it. You can stop at any time.

  • Pause any time

    Stop accepting jobs without uninstalling the app. The worker leaves the scheduling pool.

  • Schedule and idle limits

    Choose whether contribution is enabled, the idle-only wait, and active hours. The app enforces those settings locally. CPU, GPU, temperature, and bandwidth ceilings are not exposed in this build.

  • Idle-only option

    Accept work only while the PC is idle, so daily tasks keep their headroom.

  • Schedule windows

    Choose the hours the worker may accept jobs. Outside them, it stays offline.

  • Uninstall and clear

    Remove the app and its local data. The invite and enrollment token are not reused.

  • Electricity stays yours

    A working GPU or CPU draws more power. You cover the electricity; credits are not live.

Before you start

More answers in the FAQ.

Do I need a dedicated GPU?

No. LFM3-1.7B runs on CPU with 8 GB RAM and 4+ cores. A 4–7 GB GPU moves the estimate to the entry tier. CPU workers run at lower throughput and lower priority.

Can I contribute from a laptop?

Sometimes. A laptop with 8 GB RAM and 4+ cores can host LFM3, but thermals and power limits matter. The browser estimate is coarse; the native app confirms the machine.

Does the browser check certify my PC?

No. It estimates a tier from browser signals. Exact VRAM and model load are checked by the Windows app after install, and that check can revise the tier.