I am really interested in a LLM router that can track the usage and tokens used to measure the cost and put int rate limits. What this will do is help keep cost down even while using multiple LLMs.

I think this project will be a great starting point.