Ollama Local Model Setup Guide

Ollama is ideal for users with dedicated GPUs (NVIDIA or Apple Silicon) who want zero API costs and 100% offline privacy. All subtitle streams are processed entirely on your local machine.

---

1. Install & Pull Models

1. Download and install Ollama from ollama.com.
2. Pull recommended models in your terminal:

   # Blazing-fast specialized translation model
   ollama pull translategemma:4b
   
   # High-accuracy general bilingual model
   ollama pull qwen2.5:7b

---

2. Essential Step: Enable CORS

CAUTION

Browser extensions requesting http://localhost:11434 will be blocked by CORS unless Ollama is instructed to permit origins.

macOS

Launch Ollama from terminal with:

OLLAMA_ORIGINS="*" ollama serve

Linux (Systemd)

1. Edit service overrides:

   sudo systemctl edit ollama.service

2. Add the environment variable:

   [Service]
   Environment="OLLAMA_ORIGINS=*"

3. Reload and restart:

   sudo systemctl daemon-reload
   sudo systemctl restart ollama

Windows

1. Quit Ollama from the system tray.
2. Open System Environment Variables.
3. Under User Variables, add:
   • Variable name: OLLAMA_ORIGINS
   • Variable value: *
4. Relaunch Ollama.

---

3. Connecting from CoWatcher

1. Open the CoWatcher popup.
2. Select Ollama in the AI Service Provider dropdown.
3. Set Ollama API URL to http://localhost:11434.
4. Click Connect:
   • The status card will turn 🟢 Connected.
   • Model dropdowns will automatically populate with your locally installed models.
5. Select your desired Translation Model and Supervisor Model, then start watching!