Calculates and optimizes LLM response latency with TTFT, TPS, and end-to-end metrics.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.