← Software

Ollama

Local-first runtime for running large language models on Linux.

Ollama bundles llama.cpp and other inference engines behind a simple ollama run <model> interface, automatically downloading quantised open-weight models (LLaMA, Mistral, Qwen, Gemma, and many more) and serving them over a local HTTP API compatible with the OpenAI chat format.

It runs on Linux with CPU, CUDA, ROCm, or Metal backends, and has become the standard tool for running LLMs locally on a developer workstation.

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