Ethereum's Glamsterdam upgrade went live on the Sepolia test network on Tuesday, October 6, at 13:53:36 UTC (6:53 AM Pacific), right on the slot the Ethereum Foundation had scheduled. Most of the coverage led with one number, a 200 million gas limit, more than three times the 60 million that mainnet blocks carry today. An hour after the fork, though, CryptoCompass checked the chain and found the limit sitting at about 64.2 million, which left a fair question hanging: does 200 million actually happen, and how long does it take?
So we went back to Sepolia a day later and read the block headers ourselves through a public RPC node. The short answer is that the limit reached 199 million about 11 hours after the fork and has stayed there since. The longer answer says a lot about how Ethereum raises capacity, and why the testnet number isn't a mainnet promise.
Why the limit can't jump straight to 200 million
Ethereum has no switch that sets the block gas limit. Each block proposer can nudge it by at most 1/1024 of the previous block's limit, up or down, so the number only moves as fast as validators vote for it. The Foundation's Glamsterdam testnet announcement treats 200 million as a client setting rather than a consensus rule: Prysm 7.2.0 and Teku 26.9.1 supported the fork but still defaulted to 60 million, and validators had to configure 200 million by hand. Prysm 7.2.1, released the day before activation, made 200 million the default on Sepolia. The upgrade's EIP list even files the target under an informational "Gas Limit Schedule" proposal, EIP-8261, next to the networking specs.
That rule makes the fastest possible climb easy to work out. Going from 60 million to 200 million at 1/1024 per block takes ln(200 ÷ 60) ÷ ln(1 + 1/1024), or about 1,234 blocks. At one block every 12 seconds, that's roughly 4.1 hours, and only if every single proposer votes the limit up.
What the Sepolia blocks actually show
Here's the climb as recorded in Sepolia's block headers. We found the first block at or past each milestone and measured the time from the fork block's timestamp.
| Milestone | Sepolia block | Gas limit | Time (UTC) | Hours after fork |
|---|---|---|---|---|
| Fork block | 11,856,337 | 60,000,000 | Oct 6, 13:53 | 0 |
| Past 70M | 11,856,889 | 70,060,107 | Oct 6, 15:58 | 2.1 |
| Past 100M | 11,857,880 | 100,015,055 | Oct 6, 19:22 | 5.5 |
| Past 150M | 11,858,936 | 150,064,063 | Oct 6, 22:55 | 9.0 |
| Past 199M | 11,859,522 | 199,135,100 | Oct 7, 00:53 | 11.0 |
| Our latest check | 11,863,230 | 199,025,161 | Oct 7, 13:17 | 23.4 |
It took 3,185 blocks to get from the fork to 199 million, about 2.6 times the theoretical minimum of 1,234. That fits what you'd expect from gradual adoption, where some proposers had switched to the new target and others were still voting to hold at 60 million. The first hour was the slowest stretch, which matches CryptoCompass's early reading, and the pace picked up through the afternoon as more validators upgraded. Since reaching the target, the limit has hovered just under 200 million, drifting by small amounts in both directions as you'd expect when nearly everyone is aiming for the same number.
Two other numbers from the same data are worth a look. The last 300 blocks we checked used an average of 81.9 million gas against a limit of about 199.8 million, so they were around 41% full. Ethereum's fee market aims for blocks about half full, so on Sepolia there's more room than demand right now, which is normal for a test network where much of the traffic is people deliberately stress-testing it. We also counted empty slots. Of the 7,017 slots since the fork, 124 ended up without an execution block (about 1.8%), compared with 67 empty slots (about 1.0%) in the same-length window just before it. That's a small uptick, and with a brand-new block production system (enshrined proposer-builder separation) and validators upgrading at different times, it isn't surprising, but it's the kind of thing developers will want to see shrink before mainnet.
How mainnet got to 60 million
For context, Ethereum's mainnet limit sat at 30 million from 2021 until February 4, 2025, when validators raised it to 36 million. It went to 45 million on July 21, 2025, as The Block reported at the time, and to 60 million on November 25, 2025, which EIP-7935 then made the default for the Fusaka upgrade. When we checked mainnet on Wednesday morning (block 26,140,751), it was still at exactly 60,000,000. A 200 million limit would be 6.7 times the 30 million Ethereum ran on for most of its proof-of-stake life, and it would raise the network's budget from 5 million gas per second to about 16.7 million.
That doesn't mean everything gets 3.3 times cheaper or faster. Glamsterdam also reprices gas, with EIP-8037 raising the cost of creating new state and EIP-8038 updating state-access costs, so apps that write a lot of new data will get less extra room than simple activity does. The bigger blocks are only practical because of the upgrade's other headline change, block-level access lists (EIP-7928), which let clients read state and validate transactions in parallel. Our look at Vitalik Buterin's 2030 roadmap covers why that parallel work is one of the harder parts of the plan.
What it means for you
This is Geeknewz's read, not the Foundation's. If you just hold ETH, there's nothing to do; the Foundation says so plainly, and Sepolia coins have no value. If you write smart contracts, this is the week to run your code against Sepolia, because contracts that hardcode gas stipends or assume a fixed amount of remaining gas are exactly what the repricing can break. If you run a node, watch how your machine handles 200 million gas blocks on Sepolia, since bigger blocks mean more disk, memory and bandwidth.
For everyone else, the useful signal from day one is that the climb worked: clients from different teams agreed on the new rules, and validators pushed the limit to the target in about half a day without anything breaking badly. Still, Sepolia runs on a small, coordinated validator set. The next milestones to watch are a date for the Hoodi testnet, which hasn't been set, and whether client teams ship 200 million as the mainnet default or start lower. Ethereum.org lists mainnet for the fourth quarter without a confirmed date, so plan in weeks rather than days.
Data note: block figures come from Sepolia and Ethereum mainnet block headers queried through a public RPC endpoint between 13:16 and 13:19 UTC on October 7, 2026.