Setting up Ollama

Ollama allows you to run large language models locally on your own machine. Eviory OS can connect to your local Ollama instance so you can use AI completely offline without paying for API credits.

1. Configuring Ollama CORS

By default, the Ollama server rejects requests coming from an external domain (https://eviory.com). You must explicitly configure Ollama to allow https://eviory.com by setting the OLLAMA_ORIGINS environment variable.

For macOS

If you run Ollama from the terminal, run:

OLLAMA_ORIGINS="https://eviory.com" ollama serve

If you use the macOS desktop app, you need to set it globally using launchctl:

launchctl setenv OLLAMA_ORIGINS "https://eviory.com"

(Restart the Ollama app after running this command).

For Windows

1. Open the Windows Start menu and search for “Environment Variables”.
2. Click Edit the system environment variables.
3. Click the Environment Variables… button at the bottom.
4. Under “User variables”, click New….
5. Set Variable name to OLLAMA_ORIGINS.
6. Set Variable value to https://eviory.com.
7. Click OK, then restart the Ollama application.

For Linux (systemd)

Edit the Ollama service:

systemctl edit ollama.service

Add the following lines:

[Service]
Environment="OLLAMA_ORIGINS=https://eviory.com"

Then restart the service:

systemctl daemon-reload
systemctl restart ollama

2. Configuring Eviory OS Settings

Once your Ollama server is running and configured correctly:

1. Open Settings in Eviory OS.
2. Go to the AI Integrations tab.
3. Scroll down to Custom Local AI / Proxy.
4. Fill in the fields exactly as follows:

• Base URL: http://127.0.0.1:11434/v1
• API Key: Leave blank
• Model ID: Match the model you have loaded in Ollama (e.g., llama3 or phi3).

5. Click Save Changes.

(Note: If the connection fails, see the Troubleshooting CORS guide).