Ollama
Ollama focuses purely on running models and serves as a backend for other apps. If you plan to integrate local AI into projects, guides are plentiful. It runs fully offline with zero data leaks. However, it requires CLI usage and model names upfront, and large models demand high RAM.