Trace tokens, latency, and model cost
Rate limiting and monitoring for AI APIs
Framework for building AI applications