PANews reports that Vitalik Buterin has published a new article exploring the potential of the Goldreich–Kahan–Rothblum (GKR) protocol to optimize zero-knowledge (ZK) proofs. The GKR protocol utilizes a ‘batch × multi-layer’ computational structure, which significantly reduces intermediate layer commitments by solely focusing on input and output commitments. An example is provided using Poseidon2 hash as a reference. Buterin’s article outlines the recursive proof process centered around sumcheck, incorporating optimizations like Gruen’s trick, linear batching, and partial rounds with only cubic first elements. In scenarios involving polynomial commitment, these can be combined with BaseFold or FRI for further efficiency gains. Buterin asserts that the actual cost is theoretically 100 times lower compared to traditional STARK methods, potentially achieving single-digit level costs. He also cautions about the need to prevent predictability risks within circuits when employing the Fiat–Shamir challenge.