Ethereum Launches Platåberget Testnet to Prepare for Glamsterdam Upgrade

Ethereum Opens Platåberget Testnet for Glamsterdam Upgrade Testing

20, will run for several months with a publicly joinable validator set and optional client releases. This matters because several proposed changes alter assumptions currently hardcoded into transaction construction, gas estimation, and block production across the EVM stack.

Breaking the 21,000-gas invariant

The most consequential shift for application code is the separation of state access into a distinct gas dimension. Under the proposed model, account creation, contract deployment, and writes to a new storage slot are metered through a separate state-gas meter rather than folded into the standard execution gas budget.

The practical consequence: a plain ETH transfer no longer maps cleanly to a flat 21,000-gas assumption. The Foundation cautioned that this assumption no longer holds, explicitly flagging the surface where downstream tools will break. Any code path — wallets, relayers, bridges, gas oracles — that pre-allocates gas against that fixed baseline must be treated as suspect until re-validated against the new meter.

The Foundation also noted that tools relying on a hardcoded maximum gas limit will need updating, since the proposed gas repricings target a roughly 200 million gas floor. Block ceilings that worked under pre-Glamsterdam assumptions are now off the curve.

ePBS, access lists, and contract limits

Three other changes carry code-level weight. Enshrined proposer-builder separation moves auction logic into the protocol itself, removing the out-of-band PBS layer that builders and relays currently depend on. Block-level access lists alter how transactions bundle state access, changing the interface searchers and MEV infrastructure read from the chain. Larger contract and initcode limits expand what can fit in a single deployment, but also surface compatibility issues in tooling that enforces its own size ceilings.

The combined effect is that historical assumptions about block capacity, transaction packing, and builder flows no longer hold. Teams running indexers, sequencers, or custom block builders need to validate throughput models against the new ceilings before any non-finality devnet stage.

Audit checklist before Sepolia or Hoodi

Client releases remain optional on Platåberget, with temporary container images listed while teams prepare production builds. The Foundation described the testnet as an environment intended to surface downstream compatibility issues before Glamsterdam moves to longer-lived Sepolia and Hoodi stages, with mainnet activation as a later step.

For developers, the verification points are concrete:

• Replace fixed gas-limit assumptions in estimation logic; the 21,000-gas ETH transfer baseline is no longer reliable.
• Audit code paths that measure deployment cost against hardcoded initcode size limits.
• Re-validate transaction construction against block-level access lists and the proposed ~200 million gas floor.
• Confirm wallet, indexer, and validator tooling correctly parses ePBS payloads.
• Treat gas oracles and fee-payer abstractions as in-scope for regression testing against the state-gas meter.

Platåberget is a testnet notice, not a mainnet launch. But the invariants it breaks are already live assumptions in deployed code, and the surface area for silent failure is wider than the headline suggests.