Ethereum co-founder Vitalik Buterin has introduced a cryptographic protocol called GKR, short for Goldwasser–Kalai–Rothblum, to make zero-knowledge (ZK) proofs much faster and more scalable.
The protocol will lower the cost of computing ZK proofs, which are the basis for privacy and scalability solutions throughout the Ethereum ecosystem.
How GKR Speeds Up Proofs Without Heavy Commitments
In a detailed tutorial on his personal blog, Buterin described how GKR works to make computations that involve more than one data layer, like cryptographic hashing or neural network inference, more efficient.
According to him, traditional systems like STARKs require users to commit to every step in a computation, which means doing hundreds of hash operations for each byte of data. However, GKR skips these intermediate commitments entirely, instead, only committing to the inputs and outputs, which cuts down on both time and resources by a huge amount.
The core mechanism uses a mathematical process known as sumcheck. It lets the prover and verifier check complicated calculations by only looking at a few randomly chosen points instead of the whole data set.
While GKR itself doesn’t offer privacy, it can be wrapped in existing ZK-SNARK or ZK-STARK systems to provide zero-knowledge guarantees. In his tutorial, Buterin demonstrated an implementation where GKR proved millions of Poseidon2 hash functions in parallel, a structure common in blockchain operations and even in AI workloads.
The protocol’s theoretical efficiency is said to be very high, with the Ethereum co-founder estimating that it could cut overhead by 15 times compared to regular STARKs. Tests in the real world have shown improvements of just under 10x, meaning GKR-based systems could check huge cryptographic calculations for a lot less money and hardware than is required right now.
What GKR Means for Ethereum and Beyond
The introduction of GKR comes only weeks after Buterin praised the progress of the Lean Ethereum team, which is developing a minimal zkVM for long-term scalability.
His new protocol could have a big impact on how blockchain scales and how cryptographic computing works. For Ethereum, it strengthens the foundation for ZK-rollups, which are networks that batch transactions off-chain and post small proofs to the main chain, making them faster and cheaper to run.
Beyond blockchain, Buterin noted that GKR’s design suits machine learning proofs, where large models must demonstrate correct outputs without revealing the underlying data. Researchers are already looking into how GKR can be used in zk-ML inference and general-purpose proof systems.
The programmer’s recent defense of Ethereum’s 45-day unstaking queue also shows his general belief that network security and long-term trust should come before short-term convenience, a principle that GKR’s design uses to keep it both robust and scalable.