L1 zkEVM (proof-checked blocks) by Ethereum Foundation: status and timeline

L1 zkEVM (proof-checked blocks)

Let any validator confirm an Ethereum block by checking a small proof on modest hardware, so the network can carry far more transactions without forcing every node to re-execute them on bigger machines.

ProtoStage 2 of 5

Real-time proving reached in Dec 2025; optional proofs (EIP-8025) proposed for the Hegotá fork.

Updated 14 May 2026Checked 11 Oct0 updates this week

Milestones

Next · Multi-client proof devnets (Lighthouse, Prysm)

Real-time proving target defined (P99 of blocks in 10 s)10 Jul 2025Complete.
zkVMs prove 99% of mainnet blocks in under 10 s18 Dec 2025Complete.
2026 L1-zkEVM roadmap published26 Jan 2026Complete.
EIP-8025 (optional execution proofs) proposed for Hegotá14 May 2026Complete.
Multi-client proof devnets (Lighthouse, Prysm)Target 2026Current milestone.
128-bit provable security, proofs ≤300 KiBTarget Dec 2026Not yet reached.
Optional proofs on mainnet (Hegotá, after Glamsterdam)Not yet reached.
Proof verification mandatory for validatorsNot yet reached.

Most important updates

Upcoming

2026Multi-client proof devnets (Lighthouse, Prysm) (next)
Dec 2026128-bit provable security, proofs ≤300 KiB

Current obstacles

Proof systems are not yet proven secure enoughMany zkVMs rest on unproven maths. The EF set 100-bit (mid-2026) and 128-bit (end-2026) provable-security bars before relying on them.
Worst-case blocksAverage blocks prove in single-digit seconds, but an attacker can craft a block that takes minutes to prove; gas prices must be retuned to stop this.
No pay for provers yetIn the optional phase proving is altruistic: nobody is paid to make proofs, and late proofs only cost the attester timeliness rewards.

Physics limits

Proving costs far more than runningMaking a proof that a computation ran correctly takes orders of magnitude more work than running it once, so someone must still own racks of GPUs.
Security versus proof sizeHigher security margins generally mean bigger proofs, and every proof has to cross a global peer-to-peer network within seconds.
A proof only covers what the circuit saysA proof shows the zkVM followed its own code. If that code differs from Ethereum's rules, a wrong block can carry a valid proof, hence several independent zkVMs.

How it works

4 parts

Witness

Package what a block touches

An execution client outputs the block plus a 'witness': exactly the account and storage data its transactions read, so it can be checked without the full state.

Guest program

Run Ethereum's rules inside a zkVM

A standard program re-executes the block inside a zero-knowledge virtual machine, which records a trace of every step it took.

Prove

GPU provers compress the work

Provers turn that trace into a proof of a few hundred kilobytes and gossip it to other nodes over Ethereum's peer-to-peer network.

Verify

Check the proof, not the block

A 'zkAttester' checks proofs from several zkVMs. Checking takes about the same time for a 60M-gas block as for a 600M-gas one.

Spec sheet

Spec	L1 zkEVM (proof-checked blocks)
Proving time target	≤10 s for 99% of blocks^R (reported)
Prover hardware cost target	≤$100k on-premises^R (reported)
Prover power target	≤10 kW^R (reported)
Proof size target (end 2026)	≤300 KiB, no trusted setup^R (reported)
Security target (end 2026)	128-bit provable security^R (reported)
Proving time, 2025	16 min → 16 s^R (reported)
Proving cost drop, 2025	45×^R (reported)

^R reported by the company

Update log

4 updates

About Ethereum Foundation

The team behind L1 zkEVM (proof-checked blocks)

Ethereum Foundation

Ethereum research and upgrades

Swiss non-profit that funds research and coordinates upgrades to Ethereum, the largest blockchain for smart contracts. It doesn't control the network: independent software teams ship each upgrade. Its plan is to scale through rollups while keeping it cheap to run a validator at home.

The Merge (15 Sept 2022) switched Ethereum to proof of stake, cutting energy use by ~99.9%.
Fusaka (Dec 2025) lets nodes check rollup data by sampling pieces instead of downloading it all.
Glamsterdam, the next upgrade, is on test networks, with mainnet targeted for late 2026.

ethereum.foundation (opens ethereum.foundation)Company profile →

Founded: 201412 yrs
Headquarters: Switzerland
Status: Private
Valuation: private
Works in: CryptoBlockchains (L1s)
Coverage: 3 programs · 15 updateslatest 3 Sep 2026checked 11 Oct
Partners: Coinbase (Base)Eigen LabsStarkWare
People: Vitalik Buterin (opens en.wikipedia.org)Co-founder of EthereumAya Miyaguchi (opens ethereum.foundation)PresidentHsiao-Wei Wang (opens blog.ethereum.org)Co-Executive Director

Spec

L1 zkEVM (proof-checked blocks)

Proving time target

≤10 s for 99% of blocks^R (reported)

Prover hardware cost target

≤$100k on-premises^R (reported)

Prover power target

≤10 kW^R (reported)

Proof size target (end 2026)

≤300 KiB, no trusted setup^R (reported)

Security target (end 2026)

128-bit provable security^R (reported)

Proving time, 2025

16 min → 16 s^R (reported)

Proving cost drop, 2025

45×^R (reported)

L1 zkEVM (proof-checked blocks)

Milestones

Most important updates

Upcoming

Current obstacles

Physics limits

How it works

Package what a block touches

Run Ethereum's rules inside a zkVM

GPU provers compress the work

Check the proof, not the block

Spec sheet

Update log

Thu 14 May

Mon 26 Jan

Thu 18 Dec 2025

Thu 10 Jul 2025

About Ethereum Foundation

Current obstacles

Physics limits

How it works

Package what a block touches

Run Ethereum's rules inside a zkVM

GPU provers compress the work

Check the proof, not the block

Spec sheet

Update log

Thu 14 May

Mon 26 Jan

Thu 18 Dec 2025

Thu 10 Jul 2025

About Ethereum Foundation