ollama

Ollama is a tool for easily running large language models locally. It downloads and runs models like Llama, Mistral, Gemma, and Phi with a single command, making them accessible via API.

• One-command execution with `ollama run llama3` for instant model startup
• Model library covering Llama 3, Mistral, Gemma, Phi, Command R, and more
• OpenAI-compatible API for using existing OpenAI SDKs and tools directly
• GPU acceleration with NVIDIA and AMD GPU support
• Modelfile for defining custom models with Dockerfile-like syntax
• Multimodal support with vision models like LLaVA for image recognition

Built in Go for macOS, Linux, and Windows. As a ChatGPT alternative, Ollama is the definitive local LLM runtime that keeps all data on your machine. Frequently paired with Open WebUI for a complete private AI experience.