The request ledger is a searchable record of every request: model, provider, tokens, cost, latency, and status. It is the source of truth for both debugging and analytics, and the reason you do not need a separate observability pipeline to see what your LLM traffic is doing.
Why it matters
A complete, searchable record of every call is the foundation for both debugging and cost accounting - and removes the need for a separate observability pipeline.
How it works
Because the gateway sits in the request path, it records model, provider, tokens, cost, latency, and status for each call automatically. The ledger is filtered, grouped, and exported for analysis.