Ethereum puts Sepolia validators under 200 million gas blocks, 3.3 times the old ceiling

On Sepolia, the block size a validator proposes is a client setting ^[6][16]. For Prysm operators, it came down to which release they had installed. Version 7.2.0 supported the Glamsterdam fork but was built before the 200 million figure entered Sepolia's configuration. An operator who upgraded on time and changed nothing else would have kept proposing 60 million gas blocks ^[8]. Prysm shipped 7.2.1 late Monday, according to CoinDesk, with the new schedule built into the client ^[11][6]. Operators who want a different number can still set one through proposer settings, the keymanager API or Prysm's suggested gas limit option ^[16].

Each 200 million gas block holds 3.3 times the work of a 60 million one, or 140 million gas more ^[19][20]. Blocks proposed at the old default do not carry that extra load. According to crypto.news, a mix of the two sizes gives the network less consistent exposure to the larger blocks Glamsterdam is meant to test ^[9]. So the quality of the result depends on how many operators installed a release published the day before activation ^[6][2].

The run can measure only a narrow thing. A higher limit asks more of the computers that run Ethereum ^[10]. CoinDesk describes the network's step-by-step increases as a search for how far capacity can go before running a validator becomes too demanding or expensive ^[14]. Sepolia transactions use tokens with no real value ^[12], so what users pay there says little about mainnet fees. A gas limit does not set anyone's fee in any case. It adds room for transactions when demand for block space rises ^[3].

Glamsterdam, which the Ethereum Foundation places on its layer 1 scaling roadmap, also changes how blocks are produced and validated, through enshrined proposer builder separation and Block Level Access Lists ^[15]. It also changes what each fee pays for. EIP-8037 changes the cost of creating state, which the Foundation says has historically been cheap relative to the permanent storage burden it puts on nodes ^[17]. According to crypto.news, Ethereum Foundation developer Parithosh Jayanthi said during earlier Glamsterdam testing that high-level computation would get cheaper while state would get more expensive ^[18]. Fees will still depend on demand and on the kind of activity a user performs ^[4]. An application that mostly writes new storage could pay more per transaction, even in blocks with more than three times the room ^[19][18].

The test can go one of three ways. If 7.2.1 validators handle 200 million gas blocks consistently, developers move on to what CoinDesk frames as the question of how much of that capacity suits Ethereum itself ^[21]. If many operators stayed on 7.2.0 or overrode the default, there are fewer large blocks to learn from, and the result is weaker than the 200 million figure implies ^[8][9]. If validators fall behind, the mainnet step gets smaller ^[21]. I think the first number to check is how many 200 million gas blocks were actually proposed after 13:53 UTC, because that count decides how far the other readings can be trusted ^[2]. The case against that view is simple: if nearly every Prysm operator updated in time, the default never affected the data.

Mainnet is further off. The 200 million setting applies only to Sepolia ^[7], and Glamsterdam has no mainnet activation date ^[5]. If the next mainnet step follows the last one, operators will again be the ones who set it: mainnet went to 60 million gas in November 2025 after more than 513,000 validators signaled support ^[13].