OLLAMA_GUIDE.md

Using Ollama with Tandem

Tandem seamlessly integrates with Ollama to let you run powerful LLMs locally on your machine, free of charge and with complete privacy.

1. Install Ollama

macOS

1. Download Ollama from ollama.com/download/mac.
2. Install the application.
3. Open your terminal and verify installation:

   ollama --version

Linux

1. Install via the official script:

   curl -fsSL https://ollama.com/install.sh | sh

2. Verify running status:

   systemctl status ollama

   (If you want to run without systemd, follow the manual instructions on their GitHub).

Windows

1. Download the installer from ollama.com/download/windows.
2. Run the .exe file.

---

2. Pull a Model

Tandem automatically detects any model you have installed. We recommend GLM-4 or Llama 3 for a good balance of speed and intelligence.

Open your terminal and run:

# Recommended for most users (Fast & Capable)
ollama pull glm-4.7-flash:latest

# Or try Meta's Llama 3 (8B)
ollama pull llama3

Other popular models:

• mistral: Great general purpose.
• gemma: Google's open model.
• codellama: Specialized for coding tasks.

---

3. Use in Tandem

1. Open Tandem.
2. Click the Model Selector in the bottom chat bar (or go to Settings).
3. You will see an Ollama section automatically populated with your installed models.
4. Select a model and start chatting!

Note: If you install a new model while Tandem is open, just close and reopen the Model dropdown to refresh the list.