Install Ollama and Run LLMs Locally in 5 Minutes

Ollama lets you run Llama, Mistral, Gemma, and 100+ other models on your Mac, Windows, or Linux—no cloud, no API keys.

Install

macOS/Linux:

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

Windows: Download from ollama.com and run the installer.

Your first model

ollama run llama3.2

Ollama downloads the model (a few GB) and starts a chat. Type your prompt, get a response. That's it.

Other models to try

• ollama run mistral – Strong general-purpose
• ollama run codellama – Coding-focused
• ollama run phi3 – Small, fast, runs on 8GB RAM
• ollama run gemma2 – Google's open model

API access

Ollama exposes an OpenAI-compatible API on localhost:11434. Use it with any tool that supports OpenAI: Open WebUI, Continue, custom scripts. No API key needed for local use.

Hardware tips

• 7B models: 8GB RAM minimum, 16GB recommended
• 13B models: 16GB RAM, GPU helps a lot
• Apple Silicon and NVIDIA GPUs are auto-detected