Ethereum (ETH) Developers Lock Glamsterdam Sepolia Testnet Fork for Oct 6

Sepolia Fork Locked for Oct 6

Core developers confirmed on the September 17 coordination call that Glamsterdam, the next major upgrade in the Ethereum upgrade pipeline, will activate on the Sepolia testnet on October 6, 2026, at 13:53:36 UTC. The official deployment table we reviewed sets the fork epoch at 353024, with the first slot at 11296768. The corresponding rows for the Hoodi testnet and for mainnet remain marked as pending, so no mainnet date exists yet. Developers said the decision was taken only after the rehearsal results were reviewed, and the fork schedule leaves unusually little slack.

Glamsterdam is a paired release: Gloas on the consensus layer and Amsterdam on the execution layer. Its centerpiece is enshrined proposer-builder separation (ePBS), which formally splits the validator proposing a block from the specialized builders that construct it, opening block construction to a competitive bidding market. The upgrade also bundles block-level access lists, which declare in advance which state each transaction will touch — a change aimed squarely at raising block processing capacity. The date followed a completed migration rehearsal on Devnet-11, the multi-client network used for pre-release integration. That rehearsal stress-tested the network's gas fee ceiling by lifting the block gas limit from 60 million to 200 million while preserving finality throughout — a result that gave developers the confidence to commit to October 6. The margin is thin, however: the final client release deadline is September 29, leaving a seven-day validation window before the fork — half the fourteen-day minimum researcher Fredrik Svantes described as standard between a client release and a testnet going live. Sepolia is the smaller of Ethereum's public test networks, typically where client interoperability is validated before the larger Hoodi dress rehearsal, now targeted for around October 27.

Fake Builders Could Stall the Rehearsal

Alongside the date, researchers flagged an attack vector that exists precisely because ePBS will first run where money is free. Under ePBS, block-building rights are awarded through auctions; on mainnet those bids cost real capital, but on Sepolia the native test ether is distributed free. An attacker can therefore stand up thousands of spoofed builder identities, win auctions in rotation with outsized bids, and never deliver the payloads — repeatedly, at zero cost — until transaction processing on the testnet grinds to a halt. Each failed bidding round costs the attacker nothing and simply frees them to bid again.

Consensus-layer developer Potuz described the barrier to entry dismissively: one could run a thousand builders, bid high in sequence, and simply withhold every block body, adding that “any teenager” could pull it off. Developers scoped the risk carefully. No mainnet funds are exposed, because the exploit feeds on the testnet's cost-free incentives; the damage would be operational, delaying the real-environment testing that the mainnet migration depends on. ePBS itself formalizes a separation that has long existed informally through maximal extractable value outsourcing, and the testnet exercise is the first chance to see how those auctions behave under adversarial pressure. The administrative picture is less settled than the headline date implies: the mainnet upgrade and incident-response plan in the same repository is still a blank template, with the client-team coordinator and each response role unfilled. Client teams face the September 29 release deadline noted above, and the roadmap still points to a Q4 2026 mainnet window that will be fixed only after Sepolia's results are in. In practice, October 6 is the single date Ethereum has actually committed to; everything downstream is conditional. Readers tracking the market in real time can follow live spot and futures prices on MEXC.

Q4 Mainnet Hinges on Sepolia

For our desk, the two threads form one arc: Ethereum has locked its first hard date, but the lock itself is unproven. The release documentation fixes what is certain — epoch 353024, client software due September 29, activation at 13:53:36 UTC on October 6 — and node operators should treat those as the only firm calendar entries. The fake-builder warning is a testnet-economics problem, not a defect in ePBS, yet a zero-cost stalling attack would disrupt the very rehearsal the Q4 mainnet window rests on. With parallel research such as 10-second block times still advancing, and the Ethereum 1.0 to 2.0 transition as precedent for how testnet results reshape schedules, the Sepolia outcome — not the roadmap — will effectively set the mainnet date.