A draft Ethereum proposal would let transactions include off-chain STARK proofs, aiming to reduce node validation costs for complex smart accounts. The change could affect wallets, clients, and developers if adopted, but is not yet a consensus rule.
Ethereum developers are exploring a new approach to transaction validation that could shift some of the network's computational burden away from individual nodes. EIP-8361, currently a draft proposal, introduces a system where transactions can carry off-chain cryptographic proofs-specifically, STARKs-demonstrating that their validation logic has already been executed under certain state assumptions. This mechanism is designed to make it less costly for nodes to verify complex smart-account transactions before they enter the public mempool, without changing how transactions are ultimately processed on-chain.
How Proof-Carrying Transactions Work
Under the EIP-8361 proposal, a transaction built using the EIP-8141 frame format can include a succinct STARK proof as part of its peer-to-peer payload. Instead of every node simulating the transaction's validation logic, an off-chain prover runs the validation prefix once and generates a proof tied to the transaction, its payer, dependencies, and the specific state conditions assumed. When a node receives such a transaction, it checks the proof and compares the declared state assumptions against its current view of Ethereum's state. If the proof and assumptions hold, the node can admit the transaction to its mempool without repeating the full validation simulation.
This approach is particularly relevant for smart accounts with programmable validation logic-such as multisignature schemes, key rotations, or custom spending policies-that can be expensive for every node to simulate. By offloading the heavy computation to a single prover, EIP-8361 aims to keep public transaction propagation feasible even as account logic grows more complex.
Assumptions, State Changes, and Proof Lifecycle
The core of EIP-8361's design is the "assumption vector," a list of state values (like balances or storage slots) that the proof depends on. Each assumption specifies whether the current state must equal or exceed a given value. For example, a transaction might require a paymaster's balance to be at least 0.2 ETH; as long as this condition is met, the proof remains valid. If the state changes-such as a balance dropping below the threshold-nodes must recheck the assumptions and may evict or park the transaction from the mempool.
Importantly, the proof is only used for mempool admission and propagation. Once a transaction is included in a block, Ethereum's consensus rules require the full validation logic to be executed again, and the proof is discarded. It does not become part of the blockchain's permanent state, nor does it affect block validity or transaction execution. This design avoids on-chain data bloat and ensures that consensus remains unchanged.
Comparisons and Security Considerations
EIP-8361's proof system is distinct from the validity proofs used by ZK rollups, which are submitted on-chain to prove the correctness of entire batches of Layer 2 transactions. In contrast, EIP-8361 proofs are temporary, off-chain artifacts used only for mempool policy. They do not authorize state transitions or enable instant withdrawals from Layer 2 networks.
The proposal also differs from fraud-proof systems used in optimistic rollups, where invalid state transitions can be challenged after the fact. With EIP-8361, the proof must be verified before a transaction is admitted to the mempool, and there is no post-admission challenge period. The mechanism is not intended to provide privacy, guarantee post-quantum security, or serve as a scaling solution, but rather to make complex validation logic more manageable for network participants.
Risks include the potential centralization of proof generation, as producing STARKs can require specialized software and significant computation. If only a few services can efficiently generate proofs, wallets and users may become dependent on centralized infrastructure. There are also concerns about denial-of-service attacks, incomplete assumption vectors, and the need for consistent implementation across clients and wallets.
As of August 2026, EIP-8361 remains a draft and has not been adopted as an Ethereum consensus rule. Its impact will depend on further development, standardization, and community acceptance.
According to data from Etherscan, Ethereum's average daily transaction count in July 2026 exceeded 1.2 million, with smart contract interactions accounting for more than 60% of activity. As programmable account features expand, the computational cost of transaction validation has become a growing concern for node operators and wallet developers.
While EIP-8361 is not yet implemented, its approach highlights the ongoing tension between network scalability, security, and the flexibility of smart-account logic. If adopted, it could help Ethereum nodes handle more complex transactions without sacrificing performance or decentralization, but it also introduces new dependencies and operational risks that will need to be carefully managed.
