Ollama runs open-source LLMs locally. No account or internet needed after setup.
1. Install
sudo pacman -S ollama
copy
Or from AUR:
paru -S ollama-bin
copy
2. Enable and start
sudo systemctl enable --now ollama
copy
3. Pull a model
ollama pull llama3.2
copy
— 2GB, good general purpose
ollama pull mistral
copy
— 4GB, stronger reasoning
ollama pull qwen2.5:3b
copy
— 2GB, fast and lightweight
Once Ollama is running, come back here and click Test connection — the model dropdown will populate automatically.