Lagrange Coprocessor is the verification layer for off-chain computation; it’s not about running fast, but about running in a "trustworthy" manner. Through STARK proofs, it allows anyone to submit a verifiable result on-chain after completing computations off-chain. This means that only a small amount of resources is used on-chain to receive the final results of complex tasks without needing to compute them oneself.
The entire process relies on LA to maintain order: Prover and Aggregator nodes must stake LA to execute tasks, and if there is a failure in delivery or fraud, the system will automatically punish and destroy their tokens. When users pay for task fees, they also use LA as the unit of account. A portion of the consumed LA enters the incentive pool, while another portion is directly destroyed, leading to long-term deflation of the token model.
Coprocessor solves what cannot be done on-chain, and LA is an indispensable trust engine in this system.