Install on Windows

Idlerate is a single exe that runs in your user account. It does not need an administrator.

What you need

  • Windows 10 or 11 (64-bit)
  • An NVIDIA or AMD GPU (CPU-only still works with the smallest starter model)
  • An Idlerate account at relay.idlerate.dev

Install

  1. Get idlerate.exe. The Windows installer is in private preview. Sign up at relay.idlerate.dev and verify your email, and we’ll send you the download link.
  2. Check the download. Save checksums.txt from the same email next to the exe, then compare the two hashes in PowerShell:

    (Get-FileHash .\idlerate.exe -Algorithm SHA256).Hash
    Select-String idlerate.exe .\checksums.txt

    The first line must match the hash on the idlerate.exe line of checksums.txt (case doesn’t matter). Don’t run it if they differ.

  3. Open PowerShell in the folder where you saved it and run:

    .\idlerate.exe install
  4. Sign in at relay.idlerate.dev and enter the pairing code the installer prints. The same code is written to %LOCALAPPDATA%\Idlerate\pairing.txt.

That is the whole setup. The installer:

  • Copies itself to %LOCALAPPDATA%\Idlerate
  • Registers a per-user logon task named Idlerate and starts it
  • Points the PC at https://relay.idlerate.dev by default
  • Installs Ollama if it is not already running (downloads a pinned OllamaSetup.exe, checks its sha256, runs it silently; falls back to winget install Ollama.Ollama for the current user)
  • Pulls one starter model based on your GPU memory (see below)

Progress lines look like IDLERATE_PROGRESS .... The last line is IDLERATE_INSTALL_OK.

Options

.\idlerate.exe install --model llama3.2:1b   # pick the starter model yourself
.\idlerate.exe install --no-ollama           # skip Ollama and the model pull
.\idlerate.exe uninstall                     # remove the task and the exe

uninstall keeps config.json, so the machine identity survives a reinstall. It does not remove Ollama.

An MSI is also available for the same per-user install:

msiexec /i idlerate.msi

Starter model by VRAM

Available GPU memory Model pulled
Under 4 GiB (or CPU only) llama3.2:1b
4 GiB up to under 8 GiB llama3.2:3b
8 GiB up to under 14 GiB qwen2.5:7b (Q4)
14 GiB and above qwen2.5:14b (Q4)

NVIDIA and AMD both work. The gaming pause uses Windows GPU counters that cover both. After install you can add more models yourself with ollama pull <model>; they show up the next time the PC refreshes its model list.

After install

  • The tray icon (green = online, yellow = paused, gray = offline) is the local control. Use Pause, Resume, Pause while gaming, Open dashboard, View logs, and Quit.
  • Pair from the dashboard with the code. One account can pair several machines; each machine pairs on its own. See Machines.
  • Create an agent key and connect Cursor or another client. See Connect your agent.

Updating

Use the newer idlerate.exe from your download email and run .\idlerate.exe install again. Your config and pairing stay in place.