LLM Observability

Traces, latency, token usage and cost for the LLM calls your application records via the observability API. Reading these stats never counts against your quota.