Glamsterdam hits Sepolia on Oct. 6: what changes for ETH, and what doesn't
Ethereum's biggest upgrade since Fusaka goes live on a testnet this week. Mainnet isn't scheduled. Here is what ePBS, access lists and gas repricing change, and what would actually signal a mainnet date.
On Tuesday, October 6, Ethereum's next major upgrade, Glamsterdam, is scheduled to go live on the Sepolia test network. Headlines have already started treating the date as a milestone for ETH itself. So the reader question is simple: does a testnet fork change anything for people who hold or use ETH, and what would actually tell you that a mainnet upgrade is close?
The short answer
Nothing changes on Ethereum mainnet on October 6. The Ethereum Foundation's announcement says plainly that it "covers Sepolia; it does not schedule a mainnet upgrade," and that a separate announcement will cover mainnet. The Sepolia fork matters as a stress test, not as a launch. Glamsterdam rewires how blocks are proposed and built, and how gas is priced. Those are exactly the kinds of changes a public testnet is meant to break before real money is involved. The useful signals come after activation: whether Sepolia stays stable, whether a second testnet (Hoodi) gets a confirmed date, and when, or whether, a mainnet epoch is published.
What Glamsterdam actually changes
According to the Ethereum Foundation post, Glamsterdam follows the Fusaka upgrade and combines an execution-layer upgrade (Amsterdam) with a consensus-layer one (Gloas). The post lists 18 EIPs, tracked under the meta-proposal EIP-7773. Three headline changes stand out.
1. Enshrined proposer-builder separation (EIP-7732)
Today, most Ethereum blocks are assembled by specialised "builders" and handed to validators through off-protocol middleware. EIP-7732 moves that hand-off into the consensus protocol itself. In the Foundation's words, "a proposer includes a builder's commitment to an execution payload, and the builder subsequently reveals that payload," and "the protocol handles payment to the proposer, reducing reliance on trusted middleware." Reporting by Parameter (September 18) adds that this design stretches the window for distributing the block payload from roughly 2 seconds to roughly 9 seconds; that figure comes from secondary reporting, not the Foundation post.
2. Block-level access lists (EIP-7928)
Each block will carry an enforced list of the accounts and storage locations it touches, plus post-transaction state changes. The stated goal is to let clients read state from disk in parallel, validate transactions in parallel and compute state roots more efficiently. That is groundwork for raising throughput without asking node operators for dramatically more hardware.
3. Gas repricing (EIP-8037, EIP-8038)
These proposals revise what it costs to create and access state, so gas better reflects resource usage. This is the part with the most direct consequences for developers. Cointelegraph (September 29) notes that contracts relying on fixed gas stipends or hard-coded gas limits may need changes. The Foundation tells developers to test on Sepolia and follow its repricing impact guide.
The gas-limit question
CCN (September 30) describes a 200-million gas-limit target for the upgrade. The Foundation post is more granular. On Sepolia, Prysm and Teku default to a 60M gas limit after activation, and validators who want to propose with 200M must configure it explicitly. In other words, 200M is an option being exercised on a testnet, not a mainnet setting that has been decided.
Why this testnet fork is worth watching closely
Two process details make this Sepolia activation more than a formality.
A compressed review window. According to crypto.news (September 18), client releases were due September 29, only seven days before the fork. Developer Fredrik Svantes noted that the usual process calls for at least 14 days between release-ready software and the first public testnet activation. Core developer Alex Stokes urged teams to ship earlier where possible so more reviewers could examine the code. The Foundation post lists Sepolia-ready versions for six consensus clients (Grandine, Lighthouse, Lodestar, Nimbus, Prysm, Teku) and six execution clients (Besu, Ethrex, Erigon, go-ethereum, Nethermind, Reth).
A known way to grief the testnet. On the September 17 All Core Developers consensus call, developer Potuz warned that test ether is free, so an attacker could create many disposable builder identities, win auctions with inflated bids and then withhold the payloads. "I can just spin up a thousand builders, rotate them, offer very high bids, and not produce payloads," he said, as quoted by Parameter and crypto.news. Crypto Daily (September 19) stresses that this is a testnet-availability risk, not a threat to mainnet funds. Reporting indicates that builder-specific penalties were discussed, but none had been confirmed at the time of writing.
That second point is the real story. ePBS changes who is trusted to deliver a block's contents. A public testnet where bad actors cost nothing is the cheapest place to learn whether the protocol's fallback paths hold up. If Sepolia has a rough first week, that does not mean the design is broken. It does make it more likely that the Hoodi and mainnet dates slip.
What we know vs. what we're inferring
- Known (Ethereum Foundation): Sepolia activation is set for epoch 353,024, slot 11,296,768, at 13:53:36 UTC on October 6, 2026. "Hoodi and mainnet activation dates have not yet been decided."
- Known (Foundation): the headline changes are ePBS (EIP-7732), block-level access lists (EIP-7928) and gas repricing (EIP-8037/8038). Sepolia stakers must update both the beacon node and the validator client.
- Reported, not official: October 27 was tentatively discussed for Hoodi on the September 17 call (Crypto Daily, Parameter). A Q4 2026 mainnet target is cited in coverage but is not scheduled.
- Reported: Vitalik Buterin has suggested that Hegotá, the upgrade after Glamsterdam, could be Ethereum's last "normal" fork (Cointelegraph, CCN).
- Our inference: because client releases landed with half the usual review window and the builder-griefing risk is unresolved, a slip in later testnet or mainnet dates would be unsurprising. We are not predicting one.
- Our inference: any price reaction to a testnet date is speculative. Nothing about mainnet economics changes on October 6.
What to watch next
- October 6, 2026, 13:53:36 UTC (09:53 ET / 10:53 UTC-3): Glamsterdam activates on Sepolia. Watch whether the chain keeps finalising and whether client teams report bugs.
- The first days after activation: look for evidence of builders winning bids and then withholding payloads, and for how developers respond on subsequent All Core Developers calls.
- October 27, 2026 (provisional, unconfirmed): the Hoodi date discussed in September. Treat it as real only once the Foundation publishes an epoch.
- A separate mainnet announcement: no date exists yet. Until the Foundation publishes a mainnet epoch and client versions, any "Glamsterdam launch date" for mainnet is speculation.
- Developers and contract owners: check for dependencies on fixed gas assumptions now, while repricing can be tested on Sepolia.
Educational information only — not financial advice.