# How do you run a local LLM?

Install Ollama, pull a model, and chat — the fastest path to running a local LLM on Mac, Windows, or Linux.

The fastest way to run a local LLM in 2026, per Ollama's official README:

Install Ollama for your OS.
Open a terminal and run ollama pull llama3.2 (see the Llama 3.2 model page for tags).
Run ollama run llama3.2 and start chatting.
For a GUI, try LM Studio or Jan. For IDE integration, point Cursor or Continue at your local OpenAI-compatible endpoint — Ollama serves one on localhost:11434 per Ollama's API docs.

Check RAM against model size listings: 8 GB can run small 3B models; 16 GB is comfortable for 7B–8B; 32 GB+ opens 14B and some 32B quantized models.



## Sources
1. [Ollama — README](https://github.com/ollama/ollama/blob/main/README.md)
2. [Ollama — download](https://ollama.com/download)
3. [Ollama — Llama 3.2](https://ollama.com/library/llama3.2)
4. [Ollama — API documentation](https://github.com/ollama/ollama/blob/main/docs/api.md)
5. [LM Studio](https://lmstudio.ai/)
6. [Jan](https://jan.ai/)
7. [Cursor — model settings](https://docs.cursor.com/settings/models)
8. [Continue — select a model](https://docs.continue.dev/setup/select-model)

---
Page: https://localizedaiguide.com/q/how-to-run-a-local-llm