Ethereum's Glamsterdam upgrade — the largest coordinated hard fork since the September 2022 Merge — has reached public testnet. The Ethereum Foundation launched the Platåberget testnet on August 17, 2026, and core developers proposed a Sepolia fork date of September 28, 2026, at 14:44:48 UTC. Mai...
"We're working on devnets with all the EIPs in them right now. This is the last phase before we work on hardening and then shipping the testnets. There's no fixed timeline, but we've made massive progress." — Parithosh Jayanthi, Core Developer, Ethereum Foundation
Ethereum's Glamsterdam upgrade — the largest coordinated hard fork since the September 2022 Merge — has reached public testnet. The Ethereum Foundation launched the Platåberget testnet on August 17, 2026, and core developers proposed a Sepolia fork date of September 28, 2026, at 14:44:48 UTC. Mainnet activation remains unscheduled but is targeted for Q4 2026.
The fork bundles ten Ethereum Improvement Proposals under meta-EIP 7773. Two dominate the engineering effort: EIP-7732, which enshrines Proposer-Builder Separation (ePBS) directly into the protocol, and EIP-7928, which introduces Block-Level Access Lists (BALs) enabling parallel transaction execution. Together with gas repricing under EIP-8037, the changes clear a path toward a 200-million gas limit per block — a 3.3x increase from the current ~60 million — and a design target of approximately 10,000 transactions per second on Layer 1.
The upgrade carries material breaking changes for wallets, indexers, gas estimators, and a subset of deployed smart contracts. Developers who have replayed historical mainnet transactions under the new gas schedule report that most contracts are unaffected, but applications relying on hardcapped maximum gas limits will require updates.
The Platåberget testnet, announced via the Ethereum Foundation blog on August 17, 2026, is a short-term, permissionless network open to validators, builders, and application developers. Six consensus-layer clients (Lighthouse, Lodestar, Nimbus, Prysm, Teku, Grandine) and seven execution-layer clients (Besu, Geth, Erigon, Nethermind, Reth, Nimbus-EL, Ethrex) are participating. The Glamsterdam fork activated on Platåberget on August 20, 2026.
According to minutes from the developer call held on August 20, 2026, core developers proposed September 28, 2026, as the Sepolia testnet fork date, pinned to a specific epoch and slot. Six client teams confirmed the date in the session chat. The date remains provisional; formal ratification was deferred to the following All Core Developers call.
Following Sepolia, the upgrade is expected to proceed to the Hoodi testnet before mainnet. Based on the cadence of past forks, which ran two to four months of public-testnet testing, mainnet activation would fall between late November 2026 and early 2027. Core developers have stated that correctness takes priority over any specific date.
The Soldøgn interop devnet, which preceded Platåberget, concluded on May 2, 2026, after achieving stable multi-client operation.
EIP-7773 serves as the meta-EIP tracking ten proposals scheduled for inclusion:
| EIP | Description | |---|---| | 7708 | EOF-related execution changes | | 7732 | Enshrined Proposer-Builder Separation (ePBS) | | 7778 | Consensus data structure forward-compatibility | | 7843 | Additional execution-layer changes | | 7928 | Block-Level Access Lists (BALs) | | 7954 | Contract size limit increase (24 KiB → 64 KiB) | | 7976 | Initcode limit increase (48 KiB → 128 KiB) | | 7981 | Additional protocol changes | | 8024 | Additional protocol changes | | 8037 | State creation gas repricing |
The scope grew beyond initial estimates, which contributed to the timeline slipping from H1 2026 to Q4 2026. According to CoinDesk reporting from June 2026, Ethereum Foundation staff described the bundle as "the last phase before hardening."
EIP-7732 formally separates the proposer role (selecting the consensus block) from the builder role (assembling the execution payload) at the protocol level. This function currently depends on MEV-Boost, an off-chain relay system developed by Flashbots.
The existing system operates as follows: validators outsource block construction to specialized builders via trusted third-party relays. Seven major relays currently serve the network, including Flashbots, bloXroute Max Profit, bloXroute Regulated, ultrasound relay, Agnostic Relay, Aestus, and Titan. The dependency on these relays creates centralization risk and operates entirely outside Ethereum's protocol rules.
Under ePBS, every builder bid is committed on-chain. The auction process becomes publicly verifiable. Validators no longer need to trust intermediary relays to honestly transmit builder bids.
For solo stakers, ePBS provides access to MEV rewards without running complex builder software, lowering the technical barrier to competitive block proposal. For institutional operators, the standardized on-chain process simplifies validator fleet management.
Academic research has identified a residual risk: builders may strategically delay block revelation on approximately 0.82% of blocks under normal conditions, rising to an estimated 6% during high-volatility periods. Whether this theoretical vulnerability manifests in practice will depend on mainnet conditions.
EIP-7928 introduces Block-Level Access Lists (BALs), allowing blocks to declare in advance which accounts and smart-contract storage slots they will read or write. This pre-declaration enables Ethereum clients to:
According to The Defiant, ePBS, BALs, and EIP-8037 repricings combine to unlock a 200-million gas limit floor — a throughput increase that, in the Foundation's framing, "could keep network fees pinned for years." The design target is approximately 10,000 TPS, roughly 10x what the network processes today.
The change represents a structural shift in how Ethereum processes blocks. Current sequential execution means each transaction must complete before the next begins. Parallel execution allows independent transactions to process simultaneously, extracting more computational work from the same 12-second slot.
EIP-8037, finalized in May 2026, restructures how Ethereum charges for state creation — new accounts, storage slots, and deployed contract bytecode. The EIP sets a fixed cost per byte of new state and introduces a separate gas reservoir for state growth, decoupling state costs from execution costs.
The 200-million gas limit is a design target, not a value the fork itself enforces. Validators would still need to coordinate a gas limit increase via the existing signaling mechanism after the fork activates. However, Glamsterdam removes the technical barriers that previously made gas limits above 60 million impractical.
According to KuCoin reporting, the gas repricing is estimated to reduce L1 transfer fees by approximately 71%. Separately, EIP-7904, which realigns opcode costs with actual computational resource usage, is estimated to produce a 78.6% reduction in L1 fees for certain operations. Plain ETH transfers will cost more when sending to new (previously unfunded) accounts under the new state-creation pricing, a deliberate choice to charge for the long-term storage burden.
The Ethereum Foundation's repricing testing post of August 24, 2026, noted that replaying historical mainnet transactions under the new schedule flagged a "small set" of smart contracts whose assumptions shift under the new pricing, but that "the large majority of smart contracts are unaffected."
Approximately 39.7 million ETH sits in the deposit contract as of mid-2026, representing roughly one-third of circulating supply and secured by more than 1.2 million validators. Base consensus yield stands near 2.7%, with MEV-Boost adding an estimated 0.5–1.0 percentage points for validators running the relay software, pushing total returns to the 3.1–3.3% range.
Coinbase operates validators holding 4.76 million ETH — 12.16% of total staked ETH as of Q2 2026. The company connects validators to all seven major MEV relays.
Glamsterdam changes the economics in several ways:
According to Figment, enshrined proposer-builder separation "gives large allocations a more standardized, transparent process," simplifying the compliance and operational overhead for institutional staking providers.
Whether ePBS changes the total MEV captured by the system, or merely redistributes it, remains an open question. The voluntary opt-in nature of the new builder market means adoption may be gradual.
Glamsterdam's calldata repricing directly affects Layer 2 rollups that post data to Ethereum L1. Lower calldata costs reduce settlement expenses for networks including Arbitrum, Optimism, and Base, with the savings flowing through to end-user transaction fees.
The wider propagation window enabled by ePBS means Ethereum can carry more blobs per block, expanding the data-availability budget that rollups draw from. For rollups operating at capacity, this translates to higher batch throughput without additional L1 cost.
The upgrade does not change the fundamental rollup architecture, but it reduces one of the primary variable costs — L1 data posting — that determines rollup fee floors.
The Ethereum Foundation's Platåberget announcement contains an explicit warning: "Any tool that relies on a hardcapped maximum gas limit (think wallets, indexers and gas estimators) will break and needs to be updated."
Specific breaking changes include:
The Foundation's repricing testing post of August 24 confirmed that most flagged issues in legacy contracts are resolved when the gas limit increases, reducing the practical risk of contract breakage. Developers have several months of testnet operation to identify and remediate remaining issues.
Core developers are already reviewing proposals for the follow-up upgrade, Hegotá, targeted for 2027. According to Cointelegraph reporting, developers are assessing 66 EIPs for inclusion. Only FOCIL (EIP-7805), a censorship-resistance mechanism for block inclusion, is confirmed as the consensus-layer headliner.
Client teams had until September 10, 2026, to submit preference lists. Proposals lacking a clear implementation champion face automatic deferral. Frame Transactions (EIP-8141), a privacy-focused proposal, remains under consideration alongside dozens of scaling, validator, and EVM proposals.
The Glamsterdam delay to Q4 2026 has compressed the Hegotá timeline. According to CryptoSlate, the slip forced "a two-week scramble to save [the] 2027 roadmap," with developers racing to finalize Hegotá's proposed-for-inclusion list before Glamsterdam absorbs remaining engineering capacity.
Glamsterdam represents the most structurally ambitious Ethereum upgrade since the proof-of-stake transition. The fork addresses three long-standing infrastructure debts: reliance on off-chain MEV relays, sequential block execution, and mispriced state-creation costs.
The September 28 Sepolia date, if ratified, would place mainnet activation in the November–December 2026 window, assuming two to four months of testnet operation consistent with historical precedent. The scope of breaking changes — particularly around gas estimation and state pricing — suggests that ecosystem readiness, not client software stability, may determine the final activation date.
For the 1.2 million validators and $95+ billion in staked ETH, the transition to protocol-native builder auctions represents a meaningful reduction in third-party infrastructure risk. For Layer 2 operators, lower calldata costs compress fee floors. For application developers, the new gas model requires attention but is largely backward-compatible.
The data will become clearer as Platåberget and Sepolia accumulate operational history. Until mainnet activation is formally scheduled via an All Core Developers call, Q4 2026 remains a target, not a commitment.