Everything Ethereum learned. Built into one chain.
V2 rebuilds Revolution on ZKsync OS. Every batch carries a real validity proof. Accounts are smart by default. Preparation is free and settlement pays. Staking runs on a clock. And every AI agent is designed to trace back to a verified human. Open standards throughout. Nothing custom where a standard exists.
The internet is about to be run by agents. The rails were not built for them.
Automated traffic passed half of all web traffic in 2025.1 Visa, Mastercard and Ant International are aligning on how to know an agent.2 The x402 Foundation launched with 40 members to standardize payments between programs.3 Payment rails are arriving fast. Trust at the moment of settlement is not.
Meanwhile the chain Revolution launched on, EraVM, is being retired upstream.4 So we rebuilt from the ground up: on the current ZK Stack, on Ethereum's own standards, with identity and agent authority designed in from the start.
Seven layers. One chain. Every layer on a standard.
Each layer has a published specification and a stated status. Select a layer to jump to it.
Fifteen open standards and protocols. No proprietary lock-in.
Wallets, bundlers, auditors and agent frameworks already speak these. Revolution works with the tools the industry uses today.
Smart accounts through the canonical EntryPoint v0.8, at the same address as Ethereum.
Validators, executors and hooks as installable modules. Nexus accounts from an audited codebase.
The Cornerstone Paymaster follows the validation rules public bundlers require.
Rollup data published to Ethereum. Anyone can rebuild the chain from Ethereum alone.
HTTP 402 payments with a Revolution facilitator and a facets extension for verified buyers.
Intent, Cart and Payment Mandates as the canonical format. Hashes anchored on chain.
Each agent's Mandatum (delegated authority) exported as a scoped delegation any compatible wallet can issue.
A read-only mirror of every Agens (agent) identity, with transfer disabled.
An adapter over the Settlement state machine.
Merchant agents read Universal Commerce Protocol manifests to price offers.
The agent's controller key signs HTTP requests. Its .revo name travels in metadata.
Machine Payments Protocol session caps map to escrow caps in Stream mode.
Identity, proofs, Intent Book and Settlement exposed as tools and agent cards.
Parent binding, facets and the settlement record map to operator traceability and monitoring.
Existing addresses upgrade to smart accounts when ZKsync OS supports it.
The current ZK Stack. EVM equivalent. Standard Solidity, Foundry, Hardhat, viem.
01 · Chain and settlement
Real proofs. Real settlement. No shortcuts.
A rollup is only as honest as its proofs and its data. V2 gets both right.
- Every batch proven. The Airbender prover produces a validity proof for every batch from public testnet onward. The same code runs for execution and for proving.
- Data on Ethereum. V2 is a rollup that publishes to Ethereum blobs. We chose this over a validium: at our volume blob fees are small, and recoverability is total.
- Independent co-signers. Staked verifying nodes replay every block and sign each batch before commit. 2 of 3 on testnet, 3 of 5 on mainnet, with at least three mainnet signers independent of Revolution.
- Operated for the long run. Warm standby sequencer, a drilled batcher failover, keys in a KMS, versions pinned by digest.
Why it matters now
Zero-knowledge proving fell from 16 minutes to 16 seconds per Ethereum block in nine months.5 L2BEAT's first two stage checks ask whether a project has a proper proof system and data on L1.6 V2 answers yes to both.
Batch verification tolerates one offline signer on testnet. Proofs and signatures are independent checks.
02 · Accounts
Smart accounts by default. No seed phrases.
The account is the product. A Revolution account signs in with a passkey, recovers without a seed phrase, proves eligibility privately, transacts without holding gas, and grants bounded authority to agents.
- Canonical EntryPoint v0.8. Deployed at its Ethereum address, so existing bundlers, SDKs and wallets recognize Revolution with no custom setup.
- Passkeys verified on chain. The P-256 precompile is present on ZKsync OS and tested on our devnet at about 31,000 gas per verification.
- Five account classes. Personal, Professional, Business, Institutional and Agent, each with default controls. Agents hold no ownership rights.
- Curated modules. No delegatecall through sponsored paths and no arbitrary module installs in the default wallet.
Why it matters now
More than 5 billion passkeys are in active use.7 Passkey sign-ins succeed 93% of the time against 63% for other methods.8 ERC-4337 accounts have processed more than 1.29 billion user operations.9
| Class | Typical user | Default controls |
|---|---|---|
| Personal | Consumer, creator, fan | Passkey plus recovery. Sponsored cornerstone calls |
| Professional | Operator, advisor | Passkey plus hardware key. Sessions |
| Business | Company, merchant, issuer | Multisig or policy quorum. Entity verification |
| Institutional | Fund, custodian | Hardware or MPC signers. Multi-role approvals |
| Agent | AI or automation | No ownership rights. Value, target and time limits |
Whitepaper V2, section 7.2. Account classes set the baseline Potestas (permissions).
03 · Gas sponsorship
Free to prepare. Paid to settle.
A buyer will not pay to post an intent that may never fill. A fan should not pay gas to claim rewards. So every preparation step is a free cornerstone call, and only settlement pays a fee.
- A network paymaster. The Cornerstone Paymaster sponsors gas for registered cornerstone contracts only. Every target is decoded and checked.
- Funded without new issuance. The Cornerstone Budget is a share of scheduled staking emissions, 5% proposed and capped at 20% in code.
- Sybil resistant. Calls require an eligible identity, a per-identity allowance per period, and a cost cap per operation.
- Graceful when empty. If an allowance runs out, the client pays its own gas and service continues.
Built and verified
On the V2 devnet, a new account holding zero REVO made sponsored single and batch calls through EntryPoint v0.8. The budget released from emissions paid the gas. Off-target calls, ineligible senders and delegatecalls were refused in tests.
Whitepaper V2, section 7.5. Validation never reads the clock, so public bundlers accept it (ERC-7562).
04 · Programmable staking
Staking on a clock. Splits in code.
V2 keeps the published economics: 250M REVO emitted over about ten years to verifying nodes, creators and fans. It fixes how they are delivered.
- Emission by the second. Libertas (V1 network) emitted per batch. Slow batches meant it emitted about 20.8M REVO where the curve intended about 53.0M. V2 emits by wall time, so the curve holds exactly.
- Claims, not pushes. Rewards accrue through indexes and are claimed, free, by the owner, a keeper, or an agent. Cost is the same at 2 creators or 200.
- Two levels, published split. Fans stake to creator pools. Creators delegate to one node. Node fee 20% to 50%. Creator fee 0% to 100%. Self-stake carries no fee.
- Four invariants. Solvent, claimable covered, paid within emitted plus fees, cornerstone conserved. Fuzzed on every build.
Mechanics only. Revolution does not publish reward rates.
Read: Programmable staking V2 →Libertas averaged 108 seconds per batch against an assumed 40. Whitepaper V2, section 9.5.
05 · Identity
Identity for people and the agents that act for them.
Nomen (name) tells the network what you're called. Sigillum (verified identity) proves who you are. Potestas (permissions) determines what you're permitted to do. Mandatum (delegated authority) defines the authority you delegate. Agens (agent) acts within it.
- Human-bound agents. Every Agens (agent) is a sub-identity of a verified Nomen (name). It is soulbound. The parent can revoke it at any time. Agents cannot create agents.
- Verification levels. Sigillum (verified identity) L1 Basic, L2 Verified, L3 Enhanced and L3 Entity, each unlocking more.
- Prove, do not store. Over 21, in an allowed jurisdiction, accredited, sanctions clear: each is a zero-knowledge facet. The chain holds commitments, never personal data.
- Reputation you cannot buy. Records follow the verified parent. Merchant reputation is bonded with stake.
Why it matters now
A 2026 study of more than 170,000 agents in the ERC-8004 registry found a reputation can be fabricated at a median cost of $0.055 on Ethereum.10 Identities there are transferable tokens.11 Revolution binds each agent to an accountable human instead.
Whitepaper V2, sections 6.3 and 7.4. A Mandatum (delegated authority) never exceeds the parent's Potestas (permissions).
06 · ACommerce
The trust layer for agentic commerce.
At the moment of settlement, a merchant can verify that a verified human stands behind the buyer agent, that the purchase sits inside its Mandatum (delegated authority), and that the funds are real. Without learning the buyer's budget.
- One Settlement per transaction. Open, funded, delivered, settled or disputed. Funding reverts unless every required proof verifies.
- Every rail. Native stablecoin settlement through x402. Card settlements recorded and referenced. Revolution is the record of trust, not the only mover of money.
- Escrow and splits. Full, milestone and streaming escrow. Merchant, affiliate and network fee paid in the same transaction.
- Private by design. The buyer proves a purchase is authorized without revealing the budget. Price ceilings stay committed, not disclosed.
Why it matters now
Google's AP2 launched with more than 60 organizations.12 Mastercard launched Agent Pay for Machines with more than 30 partners.13 The FTC proposed a policy statement on personalized pricing in August 2026.14 Revolution interoperates with these rails and adds the trust they lack.
// HTTP 402 payment requirement, Revolution extension
{
"scheme": "exact",
"network": "revolution",
"asset": "USDC",
"amount": "4999",
"payTo": "sales.acme.revo",
"facets": ["human.verified", "age.over.18",
"jurisdiction.in:US,CA,GB"],
"requireIntentProof": true
}Whitepaper V2, section 8.6. The merchant names required facets. The buyer's Agens (agent) proves them in zero knowledge.
We tell you exactly what exists today.
Every claim on this page maps to Whitepaper V2 and its status. Here is where things stand.
- ZKsync OS devnet from one script, pinned versions
- EntryPoint v0.8 at its canonical address
- ERC-7579 Nexus accounts built from source
- Cornerstone Paymaster and Cornerstone Budget
- RevoStaking with the published split, time-based emissions
- Settlement fee routing, keeper, REST API, node agent
- 51 contract tests including 4 stateful invariants
- Zero-balance account calling a service end to end
- RNS on V2, Agent Identity Registry, Facet Registry, Proof Verifier
- Passkey validator, session executor, policy hook
- Settlement, Mandate Anchor, Escrow, Splits
- x402 facilitator with facets, proof of intent
- Reputation Registry, bonds, disputes
- Intent Book and expert affiliates
- Virtus (V2 testnet) with real proofs on Sepolia
- Verifying nodes co-signing at 2 of 3
- Public explorer and bridge portal
- Independent audits of every contract before mainnet
- Circuit audits before any facet activates
What changed, line by line.
| Area | Libertas (V1 network) | V2 |
|---|---|---|
| Stack | zkSync Era, EraVM, protocol v29 | ZKsync OS, protocol v31, EVM equivalent |
| Proofs | Placeholder verifier | Airbender validity proofs |
| Data | Described external DA, never built | Rollup, data in Ethereum blobs |
| Settlement | Running, no monitored service level | Monitored service level with alerts |
| Accounts | EraVM native abstraction | ERC-4337 v0.8 and ERC-7579, passkeys |
| Free calls | Planned protocol-level waiver | Cornerstone Paymaster, built |
| Emissions | Per batch, drifted | Per second, same curve and total |
| Payouts | Pushed during block production | Accrued and claimed, free |
| Nodes | One winning node per batch | Verifying nodes co-sign every batch |
| Names | .rev | .revo |
| Agents | Not supported | Human-bound Agens (agent) identities |
Six phases. Each one ends with a public exit test.
A phase is done when its test passes. Not when a date arrives.
-
PHASE 0
Lock down
Rotate legacy credentials. Take control of infrastructure and keys.
Exit: no working credential in any public repository
-
PHASE 1 · NOW
Foundation and Virtus (V2 testnet)
Real prover, verifying nodes, explorer, bridge, on-ramp.
Exit: a real proof executes on Sepolia; a zero-balance account claims for free
-
PHASE 2
Identity
RNS on V2, agent identities, audited facet circuits, passkeys.
Exit: agents created with a passkey and no gas
-
PHASE 3
Payments
Settlement, escrow, splits, x402 facilitator, reputation.
Exit: x402 merchants accept USDC with facet requirements
-
PHASE 4
Migration, audit, mainnet
Final snapshot, airdrop, independent audits, mainnet launch.
Exit: snapshot reconciles; audit findings closed
-
PHASE 5
Discovery
Intent Book, expert agents, agent credit, cross-chain names.
Exit: funded intents filled in the first category
One network. Five ways in.
Standard tools. One SDK.
EVM equivalent. Foundry, Hardhat, viem. SDK modules for names, identity, permissions, delegation and agents.
Developer docs →Co-sign every batch.
Staked, governance-approved verifying nodes that replay every block and send heartbeats.
Apply to operate →Pools in code.
Open a Creator Pool, set your fee, and share network rewards under published rules.
Programmable staking →Verify at settlement.
Accept agent payments with proof of a verified human, an authorized purchase and real funds.
ACommerce →Compliance-grade identity.
Accreditation, jurisdiction and sanctions checks as proofs, with data settled on Ethereum.
Talk to us →Nothing to do today. Here is what happens next.
V2 starts from a new genesis. What matters carries over by snapshot.
- Freeze. Libertas (V1 network) is frozen at a published block.
- Snapshot. Balances are exported and the snapshot is published with its hash.
- Airdrop. Balances are credited on V2, first on Virtus (V2 testnet), then mainnet.
- Names. Names move from .rev to .revo.
- Parallel run. V1 and V2 run side by side for a notice period.
- Stay safe. Revolution will never ask for your seed phrase or a fee to migrate.
Frequently asked.
01 Is Libertas still running?
Yes. Libertas (V1 network) is running today. V2 replaces it at migration, and a freeze block will be announced first.
02 What is Virtus?
Virtus (V2 testnet) is the short-lived public testnet for V2. It opens in Phase 1 with real proofs settling to Sepolia. Mainnet follows in Phase 4.
03 Why ZKsync OS?
The EraVM stack is superseded upstream. ZKsync OS is EVM equivalent, proven by Airbender, and works with standard Ethereum tooling. It removed V1's dependence on code that lived outside any repository.
04 Do the token economics change?
No. The total supply, the 250M staking emission and its 40-era curve are unchanged. V2 changes the clock from batches to seconds so the curve is delivered as designed. Revolution does not publish reward rates.
05 What are Nomen, Sigillum, Potestas, Mandatum and Agens?
They name the identity and agent stack: Nomen (name), Sigillum (verified identity), Potestas (permissions), Mandatum (delegated authority), Agens (agent). The glossary explains each one.
06 Is V2 audited?
Every contract receives an independent audit before mainnet, and circuits are audited before activation. Revolution's own V2 contracts have not been audited yet.
The full specification. Every open question listed.
Chain, accounts, identity, staking, ACommerce, economics, governance, security, compliance, migration and roadmap. Section 15.1 lists every open decision before mainnet.
Sources
- Imperva, Bad Bot Report 2026
- TNGlobal, Know-Your-Agent standards, September 2026
- Linux Foundation, x402 Foundation launch, July 2026
- ZKsync, EraVM security update, September 2026
- Ethereum Foundation, zkEVM security foundations, December 2025
- L2BEAT, Stages framework
- FIDO Alliance, State of Passkeys 2026
- FIDO Alliance, Passkey Index, October 2025
- BundleBear, ERC-4337 overview (live data)
- arXiv 2606.26028, ERC-8004 reputation study, 2026
- EIP-8004, Trustless Agents
- Google Cloud, Agent Payments Protocol
- Mastercard, Agent Pay for Machines, June 2026
- Skadden, FTC personalized pricing proposal, August 2026
Revolution facts are from Revolution Network Whitepaper v2.0. Parameters marked governance-set are proposals. This page is not an offer of any token or security.