Ollama (Local)

The 100% local mode: no data ever leaves your machine.

Prerequisites

1. Install Ollama from ollama.com
2. Download a model:

   ollama pull llama3

Launch

Method 1: Terminal (recommended for Chrome)

OLLAMA_ORIGINS="chrome-extension://*" ollama serve

The OLLAMA_ORIGINS variable allows requests from the Chrome extension (CORS).

Method 2: Permanent variable (macOS)

launchctl setenv OLLAMA_ORIGINS "chrome-extension://*"

Then launch the Ollama application normally.

Important

If you launch Ollama via the macOS app without the OLLAMA_ORIGINS variable, the extension will receive a 403 error (CORS blocked).

Configuration in the extension

1. Select Ollama in the popup
2. Verify the URL: http://localhost:11434
3. Click Save

Technical details

Parameter	Value
Default model	llama3
Endpoint	http://localhost:11434/api/chat
Authentication	None
Stream	Disabled

Recommended models

Model	Size	RAM required	Quality
llama3	8B	8 GB	Very good
gemma2:9b	9B	8 GB	Very good
llama3.2:1b	1.2B	2 GB	Decent
gemma2:2b	2.6B	4 GB	Good

To install a model:

ollama pull llama3
ollama pull gemma2:9b

Advantages

• Free: no API cost
• Private: no data sent online
• Offline: works without internet
• Customizable: use any compatible model

Troubleshooting

"Ollama is not accessible" error

• Check that Ollama is running: curl http://localhost:11434/api/tags
• Check CORS: curl -I -H "Origin: chrome-extension://test" http://localhost:11434/api/tags
• If 403: relaunch with OLLAMA_ORIGINS="chrome-extension://*"

"model not found" error

• List your models: ollama list
• Install the missing model: ollama pull llama3

Slow responses

• Larger models (8B+) require more RAM and time
• Try a lighter model (llama3.2:1b, gemma2:2b)
• On Mac with Apple Silicon chip, performance is better