why does it take 15-20 seconds to get an

becuase running a local LLM (especially when using the full context window) takes a lot of resources and can be slow depending on your hardware and what libraries you use to run the model

I expected that something that takes no time in the terminal should not take any time in the .py file

The bigger the input, the longer it takes

I doubt in the terminal you are pasting I'm several paragraphs as an input