[boarush]

Ollama is just easier to use and serve the model on a local http server. I personally use it for testing stuff with llama-index as well. Pretty useful to say the least with zero configuration issues.