ollama

Why we included this project

Running open models on your own hardware used to mean wrestling with llama.cpp directly. Ollama wraps that into a clean command-line and REST interface, so you can pull a model like Qwen or Gemma and start chatting in one command, then point your own application at the local endpoint. The REST API and official Python and JavaScript clients make it easy to drop into an existing codebase, and the Docker image gives you a clean way to run it on a server. It also handles the practical details of local inference, from model management and Modelfiles to importing custom weights. That's why it's become the default starting point for self-hosted LLM work.