Saltar al contenido principal

Ollama

Ollama lets you run large language models locally on your Mac. This document covers the basic commands to run a model and manage which ones are currently loaded in memory.

Run a Model​

ollama run qwen3:14b

This downloads the model (if it's not already available locally) and starts an interactive session with it.

Check Which Models Are Currently Running​

To see which models are currently loaded into RAM and "running" on your Mac, open a new Terminal window and run:

ollama ps

Pressing Enter shows a table with the model name, its size, and how long it has been active in memory.

Other Useful Commands​

  • List all downloaded models (running or not):
ollama list
  • Unload a model from memory immediately (force Ollama to free the RAM instead of waiting for it to unload on its own):
ollama stop qwen3:14b

Replace qwen3:14b with the name of the model you want to stop.

  • Launch OpenCode (a terminal-based coding agent) using a locally running model:
ollama launch opencode
  • Launch Claude (Anthropic's coding agent) using a locally running model:
ollama launch claude

Models​

  • qwen3:14b — general-purpose reasoning model, solid at coding. #coding #general