Glamsterdam Heads to Hoodi on 26 October With No Mainnet Date

Ethereum client teams have dated Glamsterdam's second public rehearsal. The upgrade activates on the Hoodi test network on Monday 26 October, a decision taken on this week's All Core Devs call and reported on Thursday evening and again on Friday. Releases carrying the fork are expected by Friday 16 October, which leaves node operators about ten days to install them. The exact slot sits in a pull request against the developers' own scheduling repository. Mainnet still has no date.

Why the earlier slot won

Two candidate dates went to the call and the earlier one was taken, because Devcon opens in Mumbai on 3 November, the week after. A fork landing in a conference week leaves client teams scattered when they are most likely to be needed. Not every write-up agrees on the outcome. Two published on 8 October describe Hoodi as tentatively planned for 27 October pending the Sepolia results, and a third on 6 October said no Hoodi date existed at all. The 26 October figure comes from the two accounts of the call itself; they do not reconcile, and the pull request is the thing that settles it.

Hoodi is a harder test than Sepolia

Hoodi carries staking pools and home stakers instead of the thin validator set Sepolia runs, and that is the point of forking it second. potuz, a consensus developer on the Prysm client, said as much on the call and asked for flags that would let operators switch off peer-to-peer builders and fall back to building blocks locally. The upgrade's headline change, enshrined proposer-builder separation under EIP-7732, moves block assembly to a market of outside builders.

The Sepolia run gave that market very little to look at. Only four or five builders were active there, according to the ethPandaOps monitoring team, and many blocks were still built locally anyway. September's warning about cheap builder identities turned on the opposite problem, an attacker spinning up thousands of them with free test ether. Neither condition has been exercised at scale.

The new ceiling went largely unused

Glamsterdam went live on Sepolia at 13:53:36 UTC on 6 October, at epoch 353,024, and the network finalised seven epochs later. The first Glamsterdam block carried a 60 million gas limit, matching mainnet today. About eleven hours later proposers had walked it up to 200 million, roughly 3.3 times the current mainnet ceiling and the floor the Ethereum Foundation set for Glamsterdam testing. The ramp took more than twice the four hours a 0.1 percent per block adjustment implies, partly because the Prysm and Teku clients held validators at 60 million unless operators opted in.

Capacity is not throughput. Across more than 25 consecutive blocks, usage ran between about 52 million and 92 million gas, or 26 to 46 percent of what was available, and no block came close to the limit. CoinDesk reported those gas figures early on Thursday, and the ethPandaOps explorer put 99.97 percent of eligible test stake voting toward finality.

"Highly significant. Completing this upgrade means the L1 scale will increase by at least 3x," said Barnabé Monnot, a researcher at Ethlabs.

What the mainnet schedule actually says

Here the accounts split again. One write-up on 8 October puts the mainnet target in the fourth quarter of 2026. The Ethereum Foundation's own upgrade notes, published around the Sepolia fork, say activation dates have not yet been decided and will be announced once client teams decide them. Both are on the record and they do not reconcile; the Foundation's is the primary one. The 18 EIPs in the bundle have not changed since Platåberget opened in August.

What Hoodi is being asked to prove is narrower than the gas number suggests. Sepolia showed the fork transition works and that a 200 million ceiling does not break finality on a small, professional validator set. It did not show a functioning builder market, and it did not show what home stakers do when a builder stops answering. Those are the questions the 26 October fork inherits, with client releases due ten days ahead of it.