Profiles CPU/GPU, memory, token, and runtime usage of AI model runs to estimate resource consumption and costs.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.