Get started

Run a model, flip the switch

LAPSUS is a small local app — it sits in front of the model you already run, and joins you to the community network. It bundles its own runtime, so there's nothing else to install. 

curl -fsSL https://lapsus.pyrates.io/install.sh | bash

macOS & Linux · no Gatekeeper prompt · installs to Applications and opens.
Intel Macs & Windows are on the way.

Share your AI

  1. Start LM Studio (:1234) or Ollama (:11434) and load a model.
  2. Open the LAPSUS app.
  3. Flip the Share switch on — you're part of the network.
  4. Switch it off (or quit the app) and you're disconnected instantly.

Optional: cap how much you give (e.g. 25% of idle time) and set rate limits.

Use the network

  1. Open the Use AI tab.
  2. Pick a community model, type a prompt, hit Ask.
  3. The answer comes back over a direct encrypted channel.

Remember: offer some then get some. Sharing earns credits; using the network spends them.

Download LAPSUS

One line in your terminal — macOS & Linux, no Gatekeeper prompt:

curl -fsSL https://lapsus.pyrates.io/install.sh | bash

Fetched via curl, so macOS adds no quarantine — it installs into Applications and just opens (no xattr, no “Open Anyway”). On macOS this is the way; a downloaded .zip would be blocked by Gatekeeper.

On Linux you can also grab the tarball directly:

Download for Linux (x64) →

Linux · unpack and run ./lapsus/run-lapsus.sh.

Unsigned beta. Intel Macs & Windows are on the way.

Build from source

Prefer to run it yourself? The client is fully open.

# needs Elixir + Rust (for the WebRTC NIF)
git clone https://github.com/pythononwheels/lapsus-app.git
cd lapsus-app && mix deps.get
LAPSUS_COORDINATOR_URL=wss://lapsus.pyrates.io mix lapsus.app

Full instructions are in the repository README.