Ethereum Developers Finalize Glamsterdam Gas Costs and Open Platåberget Testnet

Gas accounting on Ethereum just got a new floor. According to EtherWorld's write-up of ACDE Call #243, core developers locked the Glamsterdam gas repricing figures as final. The approved numbers cover EIP-2780, EIP-8037, and EIP-8038. Client teams no longer need to wait on a specification revision before shipping Glamsterdam-compatible builds.

The decision matters because block gas limits do not scale safely on unpriced operations. A block can pass gas accounting while forcing excessive work on every validating node. State access, memory expansion, CPU cost — each opcode class had to be measured against its worst-case execution footprint. The new figures are reported to sit inside those bounds. A few targeted benchmarks are still in flight to clients, but no spec-altering revisions are expected.

What changed in the spec

Three EIPs received finalized cost models. EIP-2780, EIP-8037, and EIP-8038 define the resource weight of operations that were previously cheaper than the node work they demanded. With these numbers locked, every client implementation can converge on the same opcode pricing without drift. The risk of divergent fork behavior across geth, besu, reth, nethermind, and Erigon drops accordingly.

Platåberget goes public

Glamsterdam Devnet-8, branded Platåberget, is the first publicly accessible devnet for this upgrade cycle. It ships with documentation and a faucet. Internal networks preceded it. Public access changes the test surface: more client diversity, more adversarial calldata, more realistic measurement of the new opcode costs against real contracts.

Devnet-7 was previously tagged as the final internal stage before a mild spec freeze. Devnet-8 extends the validation window. For contract authors, this is the first environment where the cost behavior of EIPs 2780, 8037, and 8038 can be benchmarked end-to-end against deployed code.

EIP-8261 and the 200M ceiling

Developers also backed an optional consensus-layer parameter — EIP-8261 — designed to coordinate a staged climb toward a 200 million gas limit. Activation does not jump the default on day one. Supporting clients receive a target schedule that begins near the current 60 million and steps through multiple intervals. Operators retain override authority. Clients choose independently whether to honor the schedule.

Prysm, Lighthouse, Lodestar, and Teku signaled support. Any consensus client outside that group either inherits the path or ignores it. Validator operators should already know which camp their stack sits in.

Audit checklist

Five invariants warrant direct verification before deploying against Glamsterdam-era blocks:

• Each repriced opcode charges gas proportional to its measured worst-case resource cost. Verify against the spec, not against benchmark blog posts or tooling assumptions.
• Block builders cannot assemble valid blocks whose node-side execution cost exceeds the gas-allowed budget. Stress-test with adversarial calldata, not with synthetic minimum transactions.
• EIP-8261 is optional. Confirm whether your consensus client will follow the coordinated schedule or rely on operator-set defaults.
• Deployment scripts with hardcoded gas estimates need re-evaluation. New opcode costs can invalidate assumptions baked into migration or batch-closing paths.
• Pin Platåberget as a build target now. The pre-fuzz environment is the cheapest place to find contracts that break under the new costs.