@Daniel-Liden Thanks for this great post - I appreciate the concise overview. It's been a few months now - is this still a recommended way to proxy LLM calls? And if so, where do you recommend serving the LiteLLM proxy?

If it helps, my use case is considering LangChain and I'm particularly trying to understand whether the proxy would be hosted on the same cluster running the LangChain code (potentially a hosted instance in Model Serving) or a standalone service on a separate cluster.