Ollama

Run open-source LLMs locally. Supports Llama 3, Qwen, Mistral, and 100+ models. Zero API costs, full privacy.