L1-zkEVM breakout #04, May 13, 2026

Protocol Calls & happenings
1

Meeting Time: Wednesday, May 13, 2026 at 15:00 UTC (60 minutes)

GitHub Issue

2

Meeting Summary:

This was the L1ZKVM breakout call number 4, where team members provided updates on various projects and developments. Manu presented progress on Prism, including three operational modes for proof generation and verification, with DevNet implementation of Lighthouse and Prism interop now available. Francesco shared updates on consensus layer specifications, including modifications to proof types, max proof size increases, and GPU prover support, while also noting the deprecation of validator re-signing logic. Ignacio reported on execution witness and guest program spec updates, including new releases and improvements to stateless inputs, with progress across multiple execution layer implementations including EthereX, Reth, and Nimbus. Ben discussed optimizations made to EthereX, reducing air costs by up to 26% through various improvements including hash function optimizations and removal of unnecessary dependencies. Marcin provided updates on KVM guest API standardization, highlighting five accepted standards and two new ones regarding misaligned jumps and runtime requirements, while also presenting benchmarking results on U256 operations. Han shared updates on zkoost configuration simplification and the integration of verifiers into the consensus layer, while Cody provided updates on formal verification efforts, fuzzing progress, and architecture white papers from zkVM vendors.

Click to expand detailed summary

The meeting began with participants checking audio quality, with Ladislaus confirming his connection was fine despite initial concerns about being quiet. Several participants, including Kevaundray, Luca, Francesco, and Gary, joined the meeting and confirmed they could hear each other. Gary later asked about audio quality issues, and Kevaundray clarified that the quietness was normal for everyone.

Manu presented updates on Project 4, the CL integration, focusing on three modes for Prism: proof generation, verification, and optional proof functionality. He explained that when using the optional proof feature, blocks are imported optimistically if proofs are missing or invalid, and missing proofs can be retrieved via RPC if needed. Manu also discussed the implementation of connecting to a beacon node without an execution client and highlighted the need to clean up code due to conflicts with the Prism development branch. The team is considering starting an event next week, with all details available in a specific request. Kevaundray asked about the behavior when proofs never arrive or arrive late, and Manu clarified that blocks remain optimistic until proofs are received, aligning with the current spec requirements.

Francesco provided updates on several ongoing projects, including a rewrite of the EIP with Ignacio and modifications to the consensus layer specs. He discussed changes to proof size limits, the decision to deprecate validator re-signing for DOS prevention, and updates to Beacon APIs and execution APIs. Francesco also mentioned the implementation of GPU AirProver support and the successful Lighthouse and Prism interop devnet, as well as writing a document on the Lighthouse implementation’s software architecture for potential upstreaming efforts.

The team discussed progress on execution witness and guest program specifications, with Ignacio presenting updates on releases 0.34 and upcoming changes including schema ID prefixes and chain config integration. Multiple client implementations including Ethereum, Reth, Nimbus, and Besu showed significant progress in supporting execution witness fixtures and spec compliance. The team also reviewed benchmark metrics, with improvements noted in Ethereum and Reth performance due to removal of cycle limits, and the introduction of new SIS profiling dashboard and deep branch S-load benchmarking tools.

Ben provided an update on his work with the zkVM team, focusing on optimizing guest programs. He reported a 21% reduction in gas costs through improvements to hash function calls and identified additional optimizations, including rewriting the execution witness and porting 256-bit arithmetic from Zysk. Ben is currently working on addressing crashing EST tests and welcomes further optimization suggestions from the team.

Marcin provided an update on Project 3, reporting that 5 standards have been accepted, including RISC-V target, I/O interface, cryptographic accelerators API, execution termination semantics, and memory layout restrictions. Two new standards were introduced: handling misaligned jumps and defining the runtime vendors must ship alongside ZKVMs. Marcin also presented benchmarking results on U256 operations, showing mixed performance gains with multiplication delivering strong improvements but addition and subtraction showing modest or negative results. Han shared updates on prover infrastructure, including simplified zkoost configuration and integration of verifiers into the consensus layer. Cody reported on fuzzing efforts finding bugs, formal verification work on RISC-V semantics, and progress on the SumCheck protocol, noting that NetherMind’s work has made ZISC the first zkVM to fully comply with standards except for minor gaps.

Next Steps:

  • Manu: Clean up Prism code and resolve conflicts with the Prism development branch
  • Manu: Add missing unit tests to the Prism implementation
  • Manu & Team: Consider starting a DevNet event next week with the Lighthouse and Prism interoperable implementation
  • Francesco & Ignacio: Continue iterating on the EIP rewrite PR in preparation for proposal consideration in HDR
  • Francesco: Push harder on upstreaming the Lighthouse zkVM components to the Lighthouse maintainers over the coming weeks
  • Francesco: Support the open-source contributor who offered to spec out the lib P2P proof gossip side of things
  • Francesco: Investigate and review timelines for potential alignment between optional proof dissemination and ETH P2P (raised by Gary)
  • Francesco: Investigate the current status of the execution APIs witness retrieval proposal
  • Gary: Look into the current status of ETH P2P and whether it is production-ready for proof dissemination
  • Ignacio: Cut a new release (0.35 or next version) today or tomorrow with schema ID prefix, chain config inclusion, and public keys field fixes
  • Ignacio: Review and merge the Silkworm guest program PR (first non-Rust guest program in the repo)
  • Ben: Investigate and fix crashing EST tests in ethrex
  • Ben: Cover remaining hash function calls not using pre-compiles in ethrex
  • Ben: Rewrite execution witness to use SSZ serialization instead of archive in ethrex
  • Ben: Investigate and port native 256-bit arithmetic from ZISK into ethrex if it yields improvement
  • Marcin: Conduct additional investigation on U256 acceleration interface, including bitwise operations and addition/subtraction improvements, before proposing a standard
  • Marcin: Follow up on Ruben’s concern about the read_input function from the I/O standard via a GitHub issue
  • zkVM Teams (SP1, ZISK, others): Migrate to RV64, ensure Rust SDK is no_std, and bundle I/O and accelerator interfaces into the static library
  • Kevaundray: Ask zkVM teams to opine on the Keccak-F permutation PR (vs. Keccak function) raised by Ruben
  • Han: Investigate binding of the verifier crate in different languages (e.g., Go, Python) so consensus layer clients can verify proofs locally without spinning up zkBoost
  • Han: Look into WASM compilation of the verifier library and assess versioning coordination challenges
  • Han: Get and share the size of the standalone verifier binary (without prover embedded) in response to Francesco’s question
  • Han: Finish the ZK cluster client integration in zkBoost in preparation for the on-premise cluster
  • Cody: Push ZISK RISC-V semantics formal verification work to seek external review within a couple of weeks
  • Cody: Push test results showing NetherMind/ZISK compliance with the unaligned access standard to the monitoring dashboard

Recording Access:

3

YouTube recording available: https://youtu.be/whI_KURDylw