Skip to content
Blockchain Posts

protocol

Base Beryl hardfork hits Sepolia, mainnet activation set for June 25

Beryl ships a native B20 token precompile, cuts L1 withdrawal delay from 7 to 5 days, and lands Reth V2 with ~50% lower disk and +33% throughput. Sepolia live June 18, mainnet June 25.

by 5 min read
protocolbaseethereuminfrastructure

Base's second protocol upgrade, Beryl, activated on the Sepolia testnet on June 18, 2026 at 18:00 UTC and is scheduled to activate on mainnet on June 25, 2026 at 18:00 UTC. The release ships three changes in one hardfork: a chain-native token standard called B20, a reduction of the standard L1 withdrawal delay from seven days to five, and a node upgrade to Reth V2 with sizable disk and throughput gains. Operator release notes are in base/node v1.1.0 on GitHub; the specification overview is at basehub.org/specifications/beryl-overview and the launch post at blog.base.dev/introducing-base-beryl.

What happened

The base/node v1.1.0 release, tagged June 12, 2026, marked Sepolia operators as required to upgrade before 18:00 UTC on June 18 to follow the Beryl fork. Mainnet activation lands one week later, on June 25 at the same hour. Beryl is the first Base fork after Azul. It bundles three unrelated changes that the team chose to roll together rather than ship as separate forks.

The release notes label v1.1.0 a mandatory upgrade for v1.0.x mainnet operators as well — independent of the Beryl content, the build includes a fix for a potential node halt.

What's in the fork

B20 — native token standard. Beryl introduces B20, an ERC-20 superset implemented as a node-level precompile rather than as a deployed smart contract. Tokens issued via B20 keep full ERC-20 compatibility (transfer, balanceOf, approve, permit) but expose chain-native handling for roles, transfer policies, supply caps, pausing, memos, and EIP-2612 permits. The target users named by the Base team are stablecoin issuers, real-world-asset (RWA) issuers and long-tail token operators that want richer access-control primitives without writing them in Solidity.

Withdrawal delay 7 → 5 days. A standard single-proof withdrawal from Base to L1 Ethereum currently finalises after seven days. Beryl shortens that to five. The change brings forward bridge liquidity unwinds and reduces the working-capital cost of running standard market-making and arbitrage routes that close on L1.

Reth V2 — execution-client upgrade. Base node v1.1.0 ships the V2 line of Reth, the Rust execution client maintained by Paradigm. Base's own release notes credit the V2 cut with roughly 50% lower disk usage on a full node and a state-root pipeline rebuild worth roughly +33% throughput. The release also flags peering-reliability improvements and JS-tracer performance gains.

Numbers

- Beryl activation, Sepolia : 2026-06-18 18:00 UTC (live)
- Beryl activation, Mainnet : 2026-06-25 18:00 UTC (scheduled)
- Node release              : base/node v1.1.0, tagged 2026-06-12
- Withdrawal delay change   : 7 days → 5 days (single-proof)
- B20 token standard        : ERC-20 superset, node precompile (not a contract)
- Reth upgrade              : V1 → V2.3.0
- Reth V2 disk footprint    : ~50% lower vs V1 (Base team figures)
- Reth V2 throughput        : ~+33% vs V1 (state-root pipeline rebuild)
- Source                    : github.com/base/node/releases/tag/v1.1.0
                              basehub.org/specifications/beryl-overview/

The 5-day figure is the standard single-proof window. Fast-withdrawal routes through third-party liquidity providers, which already settle in minutes by fronting the bridge, are unchanged.

What it changes for operators

  • Base node operators. v1.1.0 is mandatory before mainnet activation on June 25. Sepolia operators that missed the June 18 cutoff are already off the canonical chain. Disk savings from Reth V2 are realised on re-sync; in-place upgrades keep the V1 footprint until pruning catches up.
  • Token issuers. B20 gives RWA and stablecoin issuers a way to implement compliance primitives (pause, blocklist, allowlist, transfer hooks, memos for travel-rule data) without paying the gas premium and audit surface of a Solidity implementation. Existing ERC-20 deployments are not affected; B20 is opt-in for new issuances.
  • L1 ↔ L2 bridge users. The 5-day clock starts at withdrawal initiation. Routes built around the 7-day timing — risk-managed market-making books, treasury rebalancing schedules — get back two days of capital efficiency.
  • dApp developers. No contract changes are required. Applications that read token metadata through the standard ERC-20 interface keep working against B20 tokens. Indexers that infer custom behaviour from bytecode (e.g., proxy detection) need to extend pattern matching to the B20 precompile address.

What to watch

  1. June 25, 18:00 UTC. Mainnet activation. The first hour is the canonical reorg-risk window for any hardfork. The base/base GitHub releases and operator dashboards are the first place a stuck fork would surface.
  2. First B20 issuances. The standard is only as interesting as the issuers that adopt it. Watch for the first stablecoin or RWA token to launch as B20 rather than as a contract — the differentiator versus an audited ERC-20 deployment will be the on-chain control surface.
  3. L2Beat withdrawal data. L2Beat tracks the withdrawal window and bridge composition. The shift from 7 to 5 days will show up in the bridged-supply churn rate within two weeks.
  4. Cobalt timeline. Base has flagged its next upgrade, Cobalt, for September 2026 — native account abstraction, gas sponsorship at the protocol level, transaction batching, additional B20 functionality, and a unified node binary. Cobalt's scope is the test of whether Base is moving its OP Stack fork far enough from upstream to need its own roadmap.

Context — OP Stack divergence

Base is built on the OP Stack, but Beryl is the second consecutive Base-specific hardfork (after Azul) that ships changes upstream Optimism is not running. B20 in particular is a Base-only construct — Optimism Mainnet, OP World Chain, Mode, Zora and other OP Stack chains do not implement the precompile. The withdrawal-delay change is also a Base parameter choice rather than an upstream OP protocol change.

The pattern fits a broader 2026 trend: shared-stack L2s differentiating at the node and parameter level rather than waiting on upstream releases. Optimism's own Granite and Holocene upgrades shipped narrower changes; Base's Azul-then-Beryl cadence carries the lead chain's product priorities into its own fork. The cost is upstream-divergence risk if a future OP Stack release touches the same surfaces.

Sources:

Related stories