The Foundation is warning wallets, indexers and gas estimators that Glamsterdam's repriced gas will break any tool that hard-codes a maximum limit, and the fork lands on the new testnet within days.
The Ethereum Foundation opened Platåberget on Monday, a public testnet designed to let anyone run the coming Glamsterdam upgrade before it reaches mainnet.
Glamsterdam forks on Platåberget on 20 August, giving developers three days from launch to prep their tools. It is the first live run of enshrined proposer-builder separation (EIP-7732) and block-level access lists (EIP-7928), alongside a gas-cost overhaul that reshapes how wallets and indexers estimate transactions. The Foundation is telling any team that hard-codes a maximum gas limit that those assumptions stop working after the fork. That covers most wallets, gas estimators and blockchain indexers.
Enshrined PBS is the biggest architectural change. Roughly 88% of Ethereum blocks today are built by third-party builders and delivered through MEV-Boost relays; validators have no protocol-level fallback when a relay censors transactions or fails to deliver a payload. EIP-7732 pulls block production inside the consensus rules, using on-chain builder identities, signed bids and a new payload-timeliness committee that attests to when a block arrived. If a relay disappears mid-slot, the protocol notices without the validator needing to run a private out-of-band process.
Block-level access lists sit alongside PBS. Under EIP-7928 every block ships a manifest of the accounts and storage slots it touches and the resulting state changes. That gives light clients and stateless validation designs a cheap way to check work without replaying every transaction. It is also the load-bearing piece for raising the gas limit from around 60 million today toward 200 million, a target client teams have been testing since June.
The rest of Glamsterdam is a gas repricing that touches the '21,000 gas' floor that has priced base transactions since day one. EIP-8037 introduces a state gas dimension, and EIP-2780 and EIP-8038 decompose the cost of state access into separate charges. Contract size caps rise from 24 KiB to 64 KiB; initcode from 48 KiB to 128 KiB. Tools that estimated fees off historical multipliers of 21,000 will produce wrong answers once the new dimensions land.
Platåberget is designed to catch those failures early. Where the private devnets that carried Glamsterdam through the spring were short-lived and closed to outside operators, Platåberget was launched from mainnet's current state and made public from the start. Operators can simulate the transition on a populated chain rather than joining a network that already runs the new rules. The Foundation says the testnet will stay live for at least a few months before Glamsterdam advances to the durable Sepolia and Hoodi testnets and, eventually, mainnet.
The scope was locked earlier this year after client teams finalised Glamsterdam around enshrined PBS and parallel transaction processing. Since then the upgrade has slipped from a mid-year target into the back half of 2026, with client teams still working on how to safely raise the gas limit without breaking archive nodes. Community feedback on Platåberget is expected to run for roughly a month before the fork moves to Sepolia. The upgrade targets an average 78% fee reduction and 10,000 transactions per second once its layer-2 knock-on effects are counted.
For wallet teams the warning is blunt. Any product that stores a hardcoded 30 million or 60 million gas ceiling will present users with fee estimates that are either impossible to spend or wildly conservative. Indexers that decode blocks under existing gas assumptions will drop transactions until the schema is updated. Gas estimators that predict cost from historical block distributions will keep predicting the old distribution. None of these will surface as loud failures on the fork day. They will surface as small, silent inaccuracies that compound across the wallet UX before anyone traces the source.
That is the reason for the public testnet. Ethereum has run its last several forks through short-lived devnets that only the largest client and infrastructure teams could reasonably monitor. The Dencun upgrade shipped clean; the Pectra process was noisier and dragged. Platåberget widens the debugging surface before the fork touches user funds. Beyond Glamsterdam sits Vitalik Buterin's 'Lean Ethereum' strawmap, a multi-year rewrite that would replace the EVM entirely.
The next steps are set. Glamsterdam forks on Platåberget on Wednesday. Client software teams will spend a month collecting bug reports before proposing a Sepolia fork block. If that goes cleanly, Hoodi follows, then mainnet. If it does not, Platåberget's advantage is exactly the reason the Foundation wanted it public: everything breaks in the open, on a chain that carries no value.
