Keth - Chain-Agnostic EVM prover

Kakarot can prove blocks of Ethereum-equivalent chains, enabling any EVM chain to leverage ZK for scalability. We call this service Keth, a chain-agnostic EVM proving service with real-time proof generation. We believe this brings strong value to the broader Ethereum ecosystem: help optimistic rollups transition to ZK, reinforce the proof system of existing ZK-rollups, power ZK bridges and ZK infrastructure (ZK full nodes), and more.

Keth will become a proving powerhouse, in its ability to “STARKify” every EVM-equivalent network in the space. The first use case of Keth is to help transition every Optimistic rollup into a ZK-rollup. Secondly, Keth will allow currently expensive EVM ZK-rollups to adopt a cheaper and more performant prover stack.

Lastly, Keth will also accelerate the transition of all rollups to Stage 2 as they adopt multi-proofs in their architecture, with Keth as one of the options. Every single EVM ZK-rollup will then be able to prove its integrity using multiple proof systems simultaneously without having to pay extremely high proving fees (e.g., Keth, Zeth, Succinct’s Reth, TEE etc. all at the same time). These multiple proof systems will all agree on the state of network, ensuring the security of the value transfer on top of it with no single point of failure.

For example, Taiko, an existing EVM ZK-rollup can adopt multi-proof approach with its prover adaptor, Raiko using proof systems from Keth, Zeth and Succinct’s SP1 Reth to finalize transactions.

We believe that Keth is contributing towards a “ZK-everything” future where all bridges, wallets, or full nodes in the rollup ecosystem are powered by ZK. This movement will significantly reduce the cost of every major rollup infrastructure while increasing users’ sovereignty (i.e. users gain the ability to verify proofs locally if they want, reducing the reliance on trusted infrastructure providers).

When it comes to performance, early benchmarks have shown that Starkware STWO, the prover used by Keth, achieves a substantial improvement relative to the current generation of provers in the space (benchmarks recorded in MHz rather than kHz). More accurately, STWO, which is still in early developments, could achieve tens of MHz, while the current state-of-the-art prover, RISCZero’s zkVM 1.0 could achieve up to 1MHz on a strong GPU. Keth is believed to bring exponentially improved performance, orders of magnitude better than the incumbents.

With this orders-of-magnitude better performance, we expect Keth to become the most advanced and performant EVM proving engine in the sector.

Keth is scheduled to go live in production by Q3 2025.