How to Run Llama 3.2 Locally on Windows with Ollama

Private Local Artificial Intelligence: Installing Ollama on Windows

Running artificial intelligence models locally on your personal computer guarantees total data privacy, offline availability, and zero API subscription fees. Ollama has emerged as the premier open-source tool for running large language models (such as Llama 3.2, Mistral, and Qwen) directly on Windows PCs. At Android People (androidpeople.in), we walk you through the complete setup process.

Step 1: Downloading and Installing Ollama Desktop

Visit the official download portal at ollama.com and click the Download for Windows button to save the OllamaSetup.exe installer. Run the installer to set up the core background service in your user directory. Once installed, Ollama automatically registers a system tray icon, confirming the local AI server is active.

Step 2: Pulling and Chatting with Llama 3.2 in Terminal

Open Windows PowerShell or Command Prompt and execute your first model pull command:

ollama run llama3.2

Ollama will automatically download the 2GB model weights, detect available GPU acceleration (Nvidia CUDA or AMD ROCm), and launch an interactive chat interface directly inside your command terminal.

Customizing Model Storage Location to Free C Drive Space

By default, downloaded AI model weights are saved in your user profile on the C: drive. If you want to store large model weights on a secondary drive (e.g. D:\OllamaModels), follow these steps:

1. Open Windows System Properties and click Environment Variables.
2. Under System Variables, click New.
3. Set Variable Name to OLLAMA_MODELS and Variable Value to D:\OllamaModels.
4. Click OK and restart Ollama from your system tray.

Frequently Asked Questions

Do I need a high-end graphics card to run Llama 3.2?
Llama 3.2 (3B) runs smoothly on CPUs with 8GB RAM, but an Nvidia or AMD GPU significantly accelerates generation speed.

For more software tutorials and local AI guides, keep visiting Android People.