getting started
Run OCC Node
OCC Node is the local runtime. It runs a small language model on your hardware, retrieves verified knowledge from the server privately, and connects to the broker network. Setup takes a few minutes.
Prerequisites
- —Python 3.11 or newer — download and run the installer. On Windows, check Add Python to PATH before clicking Install.
- —Git — download and run the installer, all defaults are fine.
- —Ollama — OCC checks for it automatically and will prompt you if it is missing.
Installation
Open a terminal, then clone the repository:
git clone https://github.com/VikFinlay/occ.git
cd occThen run the setup script for your platform:
Windows
start.batmacOS / Linux
./start.shThe script walks you through setup step by step — it checks all requirements, installs dependencies, and launches OCC automatically. At the end, it creates an OCC Node shortcut on your desktop so future launches require no terminal at all.
First launch
On the first run, OCC detects your available VRAM and downloads the appropriate model automatically. The table below shows which model you will get. This download only happens once — subsequent launches start in seconds.
All tiers use the Qwen family with adaptive context optimised for your hardware, and support multimodal input. Learn more →
Launching OCC
Double-click the OCC Node icon on your desktop. The server starts and OCC opens in your browser automatically — no terminal needed.
Manual fallback
If the shortcut does not work, open a terminal in the OCC folder and run:
python node/apps/gui/server.pyThen open http://localhost:7891 in your browser.
Updating
Inside OCC Node, open Settings and click Update to Latest. OCC will pull the newest version, reinstall dependencies, and restart automatically.
Manual fallback
If the button does not work, open a terminal in the OCC folder and run:
git pullCommands
/clearClear conversation history and reset context
! <query>Force network mode — consult peers regardless of local knowledge
/packsList all loaded packs and domains
/peersShow active peer nodes on the broker
/statusShow current config
/local onUse local packs only — for private Forge packs
/local offUse server packs (default)
/unloadUnload model from VRAM
/loadReload model into VRAM
/openrouter onSwitch to OpenRouter (if configured)
/openrouter offSwitch to local model
Need help?
Open an issue on GitHub or start a discussion in the repository.