Ollama + Cognito
Set up Ollama so the Cognito Chrome extension can use your local AI models — completely private, running on your machine.
Ollama runs a local server at http://127.0.0.1:11434. The guides below configure your system so the Cognito extension can talk to it.
Before you start
- Install Ollama from ollama.com. Open the app or run
ollama serve, then pull a model (e.g.ollama pull llama3.2). - Install the Cognito extension in Chrome (or any Chromium-based browser like Edge, Brave, or Arc).
Why is setup needed?
By default, browsers prevent extensions from connecting to services running on your computer. This is a security feature.
Ollama has a setting called OLLAMA_ORIGINS that controls which apps are allowed to connect. Without it, the extension will show a "connection failed" or "CORS" error when trying to reach Ollama.
The setup takes under 2 minutes and you only need to do it once.
Choose your operating system
Follow the guide for your OS. Each guide walks you through setting OLLAMA_ORIGINS step by step.
| Platform | Guide | Time |
|---|---|---|
| Windows | Setup on Windows | ~2 min |
| macOS | Setup on macOS | ~2 min |
| Linux | Setup on Linux | ~2 min |
Ollama Cloud models
Want to use larger models that don't fit on your machine? Ollama Cloud models (e.g. gpt-oss:20b-cloud, gpt-oss:120b-cloud) run on Ollama's servers. You still use them through your local Ollama — just sign in at ollama.com and link your device.
| Scenario | Guide |
|---|---|
| Ollama Cloud models | Setup Ollama Cloud |
After setup
- Open the Cognito extension in Chrome.
- Choose Ollama as the AI provider.
- Pick a model from the dropdown (e.g.
llama3.2). - Start chatting.
You can add more models anytime from the extension's Ollama panel or by running ollama pull <model> in your terminal.