# Prysm ships a last-minute fix so Ethereum's Sepolia testnet can run the Glamsterdam upgrade at 200 million gas per block

Without version 7.2.1, validators that had not changed their settings by hand would have kept proposing 60 million-gas blocks after today's 13:53:36 UTC fork. That would have weakened the capacity test that will guide mainnet limits.

- Published: 2026-10-06T05:32:21.128Z
- Canonical: https://polylog.news/crypto/2026-10-06/prysm-ships-a-last-minute-fix-so-ethereum-s-sepolia-testnet
- Publisher: Polylog (Crypto desk)
- Section: crypto
- Sources: [CoinDesk](https://www.coindesk.com/tech/2026/10/06/ethereum-s-glamsterdam-test-gets-last-minute-fix-before-major-capacity-jump), [ChainSafe Lodestar (GitHub)](https://github.com/ChainSafe/lodestar/pull/10172), [Kiln](https://www.kiln.fi/post/glamsterdam-ethereums-next-hard-fork-explained), [GetBlock](https://getblock.io/blog/ethereum-glamsterdam--what-we-know-so-far-about-the-hard-fork)

Prysm, one of Ethereum's main consensus-layer validator clients, [released version 7.2.1 on Monday evening](https://www.coindesk.com/tech/2026/10/06/ethereum-s-glamsterdam-test-gets-last-minute-fix-before-major-capacity-jump), a few hours before the Glamsterdam upgrade activates on the Sepolia test network. The previous release came out before Sepolia's parameters changed. Validators on that version who had not edited their settings would have kept building blocks at the old 60 million gas limit. With 7.2.1, they will automatically propose 200 million-gas blocks once the fork activates at 13:53:36 UTC on October 6. Other client teams made matching changes. Lodestar, for example, [merged a change scheduling the 200 million gas limit](https://github.com/ChainSafe/lodestar/pull/10172) for Glamsterdam on Sepolia.

Gas measures the computation a block may contain, so the test more than triples the work each Sepolia block can hold. Developers want to know whether validators can reliably receive, check and attest to blocks of that size before they choose a limit for the main network.

Glamsterdam has two headline changes. The first is EIP-7732, the Ethereum Improvement Proposal (EIP) for [enshrined proposer-builder separation](https://www.kiln.fi/post/glamsterdam-ethereums-next-hard-fork-explained). That change builds into the protocol the split between the validator who proposes a block and the specialist who assembles it. Today that split depends on external relays. The second is EIP-7928, which adds block-level access lists. Each block will carry a list of every account and storage slot it touches. Clients can then load state in advance and execute non-conflicting transactions in parallel. A set of gas-repricing proposals accompanies both. Together they are meant to make a [target of about 200 million gas](https://www.kiln.fi/post/glamsterdam-ethereums-next-hard-fork-explained), roughly 3.3 times today's mainnet limit, safe to run.

The mainnet date is not set. Earlier plans aimed at the third quarter, and [outside estimates have moved to the fourth quarter of 2026](https://getblock.io/blog/ethereum-glamsterdam--what-we-know-so-far-about-the-hard-fork) or later. Configuration problems like the one Prysm fixed have caused trouble before. In February 2025, misconfigured deposit-contract settings in several execution clients left the Holesky testnet unable to finalize blocks for an extended period during the Pectra rehearsal.

The upgrade affects where activity on Ethereum takes place. Ethereum holds $54.13 billion of the $96.50 billion in total decentralized-finance (DeFi) value tracked by DeFiLlama. Layer-2 networks, which process transactions off the main chain and settle results back to it, secure another $51.29 billion according to L2Beat. Base leads at $16.37 billion, and Arbitrum One holds $11.49 billion. A main chain with roughly three times the capacity competes directly with those rollups for ordinary transactions.

## What this means

More block space on the main chain lowers what each transaction pays for scarce capacity. That reduces fee revenue and ether burned per transaction, unless demand rises to fill the space. It also weakens the reason to settle activity on rollups such as Base and Arbitrum. The cost falls on node operators. Larger blocks raise bandwidth, storage and validation-time requirements, and home validators feel that first. The decentralization question is therefore whether the network keeps its smallest operators. Most coverage treats the gas increase as settled. In fact, Sepolia tests whether the parallel-execution and access-list machinery reduces the slowest validators' workload as much as designed. The desk's forward read is that if Sepolia runs at 200 million gas for several weeks without a rise in missed slots or delayed finality, client teams will set a mainnet date and raise the limit in stages over the following months rather than in one jump. The read would prove wrong if Sepolia shows validators falling behind on large blocks. That would push developers toward a lower interim target and delay mainnet into 2027.

## What to watch

- Sepolia's rate of missed slots and its time to finality in the first days after the fork. That data shows directly whether ordinary validators can handle 200 million-gas blocks.
- Whether core developers set a mainnet activation date on their next calls. A date would show that the test passed in their judgment.
- Fee levels and transaction counts on Base and Arbitrum after any mainnet capacity increase, since cheaper main-chain space competes with them directly.