Base is shipping its biggest token-layer change yet, and the timeline is fixed: Beryl activates on Base Mainnet at a single, pre-set moment. Whether you hold tokens, build on the chain, or run nodes on Base, here is what the upgrade changes and what it does not.
What Beryl activates on June 25 at 18:00 UTC
Beryl is a scheduled Base Chain hardfork (not a token launch) that activates on Base Mainnet at Unix timestamp 1782410400, or June 25, 2026 at 18:00 UTC . It carries exactly three confirmed changes. First, it introduces B20, Base's native token standard implemented as Rust precompiles rather than EVM contracts. Second, it cuts the single-proof withdrawal finalization window from 7 days to 5 days, while the dual-proof fast path stays at 1 day. Third, it ships Reth V2, which Base claims delivers up to 50% disk reduction and +33% throughput .
Base Sepolia ran the same upgrade on June 18, 2026 at 18:00 UTC as a testnet dry run . Mainnet behavior, activation success, and the performance figures cannot be verified until after the June 25 cutover. Treat all production claims as scheduled, not proven.
| Change | Before | After | Who it affects |
|---|---|---|---|
| B20 token standard | ERC-20 smart contracts only | Native Rust-precompile tokens with ERC-20 selector parity | Issuers, wallets, integrators |
| Single-proof withdrawal window | 7 days | 5 days (dual-proof stays 1 day) | Users bridging to Ethereum L1 |
| Reth V2 execution client | Prior Reth build | Up to 50% less disk, +33% throughput (claimed) | Node operators |
Note what is not in Beryl. Flashblock Access Lists, Glamsterdam EIPs, and a single client binary appeared in Base's earlier Azul roadmap, but the current Beryl overview and v1.1.1 release notes document only B20, the withdrawal-timing change, and Reth V2 . Do not assume those roadmap items ship on June 25.
B20: ERC-20 selector parity, protocol-level policy gates
B20 is Base's native token standard, built as Rust precompiles rather than EVM smart contracts. That architecture accounts for why it looks familiar but behaves differently at the edges. Tokens are deployed through a singleton B20Factory precompile whose address starts 0xB20f, while the tokens it creates start 0xB200 . Addresses are deterministic, derived from createB20(variant, salt, params, initCalls), with the variant encoded into the address itself .
For wallets and apps, the headline is selector parity. B20 preserves the full ERC-20 interface. The same transfer, transferFrom, approve, balanceOf and permit-style flows all behave as expected, so existing integrations need no rewrite . The break appears at the edges: a PolicyRegistry BLOCKLIST or ALLOWLIST can fire a PolicyForbids revert that a plain ERC-20 would never produce. Base's payments guide tells integrators to surface these reverts and simulate transfers before signing .
There are two variants, each tuned to a different issuer:
- Asset: configurable decimals from 6 to 18, a rebase multiplier, batch minting, onchain announcements, and an added
OPERATOR_ROLE. - Stablecoin: decimals fixed at 6 plus an immutable, self-declared fiat code such as USD or EUR. Base notes this code is not verified against any external registry .
The target users are stablecoin issuers, real-world-asset and equity issuers, and long-tail token creators . With B20, compliance controls sit at the chain level. Roles like MINT_ROLE, BURN_ROLE, PAUSE_ROLE and freeze-and-seize controls are baked into the chain, not into a contract the issuer deploys and governs alone .
That shift raises questions for regulated issuers. As of writing, no stablecoin or RWA issuer has publicly stated whether protocol-level freeze-and-seize is an adoption driver or a centralization tradeoff versus status-quo ERC-20 contracts. A freeze enforced by the chain is harder to dispute than one written into an issuer-controlled contract, which can read as either stronger assurance or reduced issuer sovereignty. SpotedCrypto has requested comment from regulated issuers and will update this section with attributed responses.
Who needs to act before June 25 18:00 UTC
Node operators face the highest urgency before Beryl's mainnet activation. Base requires every mainnet operator to upgrade their execution and consensus clients to base-reth-node v1.1.1+, base-consensus v1.1.1+, or base/node v1.1.1+ ahead of the cutover . The GitHub v1.1.1 release notes use explicit "must upgrade" language and confirm the release adds Beryl mainnet support across both clients . An un-upgraded node risks falling out of consensus at activation.
A few other groups should check their setup before the cutover:
- Developers building with B20. Stock Foundry cannot simulate calls to precompile addresses that hold no contract bytecode, so the required toolchain is Base's Foundry build, installed via
base-foundryup --install v1.1.0. Before mainnet, B20-active environments are Base Sepolia (chain ID 84532), Vibenet (chain ID 84538453), and local base-anvil (chain ID 31337) . - Integrators accepting B20 tokens. Implement pre-send simulation so users see a reason before they sign. Base's payments guide instructs integrators to surface B20-specific reverts such as
PolicyForbidsand paused-transfer errors . These errors do not exist in plain ERC-20 and represent a live integration risk: a transfer that passes every ERC-20 check can still revert at the policy gate. - Traders and token holders. No action required. ERC-20 selector parity means wallet compatibility is preserved, and the withdrawal path improves modestly, with the single-proof finalization window dropping from 7 days to 5 days .
Infrastructure and integration teams carry the pre-launch burden; end users inherit the upside without a migration step.
What to watch after activation
The first test of Beryl is whether its claims hold under production load. Base says B20's Rust precompiles are faster and cheaper than ERC-20 contracts, and that Reth V2 delivers up to 50% disk reduction and +33% throughput . None of those numbers can be confirmed until after the scheduled mainnet activation at 2026-06-25 18:00:00 UTC . Only on-chain data will settle the question.
Four signals are worth tracking in the first 48 hours after cutover:
- Gas benchmarks. Base has published no exact figures for B20 transfers versus ERC-20 contracts, only qualitative claims . Compare real transfer costs once tokens deploy.
- Issuer adoption. The policy model targets stablecoin, RWA, and equity issuers . Watch whether first movers adopt protocol-level compliance or keep audited ERC-20 patterns for governance flexibility.
- Reth V2 metrics. Monitor Base's dashboards and node operator reports for whether disk and throughput gains materialize .
- Missing roadmap items. Flashblock Access Lists, Glamsterdam EIPs, and a single client binary appeared in the earlier Azul roadmap but are not documented as part of Beryl . Expect a follow-on upgrade naming their milestone.
Beryl ships a real architecture change, but every performance and compliance claim is provisional until post-launch data comes in. The next 48 hours of on-chain metrics will show whether the numbers hold.
Frequently asked questions
What is B20, and how does it differ from ERC-20?
B20 is Base's native token standard, described by Base as its "own version of ERC-20" but implemented as Rust precompiles rather than EVM smart contracts . It preserves full ERC-20 selector parity: the same transfer, transferFrom, approve, balanceOf, and permit signatures, so wallets and apps keep working unchanged. The difference is enforcement: B20 moves issuer controls like blocklists, allowlists, and freeze-and-seize into the chain itself, so a transfer can revert where a plain ERC-20 would succeed .
Do node operators need to upgrade before June 25?
Yes. Base requires mainnet operators to upgrade to base-reth-node v1.1.1+, base-consensus v1.1.1+, or base/node v1.1.1+ before Beryl activates at 18:00 UTC on 2026-06-25 . The v1.1.1 release notes use mandatory language (operators "must upgrade") and state the release adds Beryl mainnet support to both clients . Nodes left on older versions risk falling out of consensus at activation.
Will existing wallets and dapps break after Beryl?
No, not for standard ERC-20 tokens. There is no change to how they behave. B20 tokens are ERC-20 selector-compatible, so wallets display and sign them through normal flows . The exception is integrators accepting B20 as payment: Base's integration guide instructs them to add pre-send simulation to surface B20-specific reverts such as PolicyForbids and paused-transfer errors before a user signs .
What is the difference between B20 Asset and B20 Stablecoin?
The Asset variant supports configurable decimals from 6 to 18, a rebase multiplier, batch minting, an added OPERATOR_ROLE, and extra metadata . The Stablecoin variant fixes decimals at 6 and adds an immutable, self-declared fiat currency code such as USD or EUR. Base explicitly notes this code is not verified against any external registry: it is a label the issuer asserts, not an external attestation .
Does Beryl ship Flashblock Access Lists and Glamsterdam EIPs?
Not confirmed. Those features appeared in Base's earlier Azul roadmap but are absent from the current Beryl upgrade specs and v1.1.1 release notes . The documented Beryl scope is three items: the B20 token standard, a reduction of the single-proof withdrawal window from 7 days to 5 days, and Reth V2 . Treat any wider feature list as a future upgrade until Base documents it.
