Summary: Gevulot is best understood not as a general-purpose smart-contract chain, but as a proving-specialized coordination layer that puts prover and verifier orchestration inside its own blockchain. Its core mechanism is a layer-1 network where users submit proving workloads, validators order them into blocks, provers are assigned jobs via VRF-based allocation, and randomly selected prover-verifiers attest to correctness before rewards are distributed. That makes Gevulot a useful comparison class for proving-market infrastructure because it pushes more of the workload-routing, verification, and reward logic into a dedicated chain instead of treating proving as an offchain market attached to another execution environment.
What it does:
Lets users deploy and run zero-knowledge prover and verifier programs as onchain programs on a purpose-built network
Uses validators to order workload transactions and maintain replicated state, while separate prover sets handle proof generation and proof verification
Supports user-specified workload parameters such as resource requirements and maximum compute time, which drive fee calculation and scheduling
Uses VRF-based workload allocation, fallback provers, and verifier subsets to distribute jobs and confirm proof validity
Distinguishes between the currently available permissioned Firestarter prover network and the planned permissionless ZkCloud architecture described in the docs
Positions itself as infrastructure for making zk-based systems easier to deploy by treating proof systems more like deployable programs than bespoke prover pipelines
Key claims:
The docs introduction calls Gevulot the first universal proving layer for ZK and says its core value proposition is low cost, high performance, high liveness and availability, permissionlessness, and credible neutrality
The main repo README describes Gevulot as a permissionless and programmable layer one blockchain for deploying zero-knowledge provers and verifiers as onchain programs
The ZkCloud overview says the network aims for centralized-equivalent proving performance while preserving decentralized-network liveness and censorship-resistance through a dual-node architecture of provers and validators
The network-actors docs show that Gevulot is not just matching jobs to hardware; it also defines staking, prover-set admission, proof-of-workload qualification, fallback allocation, and verifier selection as protocol-level control surfaces
The proving-workloads docs make clear that proof verification and reward distribution are embedded in the chain workflow: workloads are submitted as transactions, allocated after finality, verified by randomly selected provers, and then finalized in subsequent blocks
The FAQ argues Gevulot needs to be a layer one because inheriting another chain’s cost structure and performance would make it uncompetitive with centralized provers, which is an unusually explicit statement about where the team thinks the real bottleneck sits
The current README also notes that the original devnet implementation has been phased out in favor of Firestarter and the upcoming ZkCloud architecture, which matters because the live product story and the legacy node repository are no longer the same thing
Whitepaper: No canonical public whitepaper or litepaper was surfaced in the official docs during this pass. The strongest primary materials were the official docs and repository README; see ../whitepapers/gevulot-primary-sources-2026-05-08.md.