Parses a reference guide to LLM benchmarks like MMLU, HumanEval, GSM8K, and LMSYS Arena.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.