Ethereum Sepolia Forks to Gloas
Ethereum activated the Gloas consensus layer fork in the Glamsterdam upgrade on the Sepolia testnet at 13:53:36 UTC on October 6, 2026, signaling a gas limit of 200 million. The corresponding epoch is 353024, and the slot is 11296768.
The Glamsterdam upgrade combines the execution layer Amsterdam and the consensus layer Gloas in a single public test. Planned changes include the separation of proposers and builders, as well as a block-level access list to support higher throughput. Node operators are required to upgrade their clients by October 5. Gloas also introduces a builder interface, a second version of proposer settings, remote signing, and builder circuit breakers.
The 200 million gas limit is not a mainnet parameter and is not a hard rule automatically executed by all validators. The gas limit on Sepolia before the fork was approximately 60 million. Prysm 7.2.0 was released before the configuration announcement, with a default still set at 60 million; operators must manually change it to 200 million using the second version of proposer settings or keymanager interface. The old suggested-gas-limit is only valid for builders registered before Gloas.
Hours before the fork, Prysm released 7.2.1, incorporating the 200 million into the default proposal. CoinDesk reported that this was to prevent some validators from continuing to produce low-capacity blocks. The documentation for Teku 26.9.1 still states the default is 60 million unless operators add validators-builder-registration-default-gas-limit=200000000. Therefore, the test does not guarantee that the entire network will actually reach 200 million.
There are no scheduled dates for the Hoodi testnet or the Ethereum mainnet. There are no changes in supply, nor new figures for fees or burn rules. Issues such as missed upgrades, client incompatibility, or mixed configurations of 60 million and 200 million will affect block production on the testnet, not mainnet settlement.
This is a capacity rehearsal, not a mainnet transaction event. The sellers are validators' clients that remain at the default value of 60 million, while the buyers are applications and builders that want to fit more computation into layer one. The event is driven by the timing of the fork, not by price. Beneficiaries are operators who complete the upgrade and can propose 200 million gas blocks. Those under pressure are unupgraded nodes and expectations for scaling that have not yet received dates for Hoodi and the mainnet.
Source: Public Information
ABAB AI Insight
The path for Ethereum's scaling is to first raise the gas limit from about 15 million to 30 million, then to 60 million, with each increase relying on validator votes rather than a one-time hard limit. Glamsterdam moves the next step to the testnet: combining the execution layer Amsterdam with the consensus layer Gloas, with Sepolia testing 200 million at epoch 353024. The mainnet and Hoodi have no dates, indicating that capacity figures precede activation commitments.
Resource mobilization is in the client, not in funds. Prysm 7.2.0 was released just before the parameter announcement, defaulting to 60 million. 7.2.1 incorporated 200 million into the default proposal hours before the fork. Teku 26.9.1 still requires operators to manually add parameters. The motivation is to prevent half of the validators from producing low-capacity blocks, which could lead to false negatives in test results. The separation of proposers and builders, along with block-level access lists, aims to decouple block production and validation, allowing larger blocks to propagate.
A parallel can be drawn with Dencun in 2024, which will first run blobs on the testnet before moving to the mainnet, similar to Bitcoin's approach of leaving block capacity disputes in client default values. The current situation is a rehearsal, not mainnet control. The 200 million limit only applies to Sepolia and depends on who upgrades to 7.2.1 and who remains at 60 million.
Structurally, this is a preliminary step towards technical substitution. If layer one capacity moves from 60 million to 200 million, computations that originally had to be placed on layer two can remain on layer one. The mechanism is that the gas limit is jointly determined by proposer preferences and client default values, not by the name of the fork. A version difference in default values can result in a threefold difference in outcomes. The limit is determined by those producing blocks, not by the name.
The capacity of the testnet is not a commitment for the mainnet.
A version difference in default parameters results in a threefold difference in outcomes.
The limit is determined by those producing blocks, not by the name.