Ethereum researchers file a plan to cut finality from about 16 minutes to four, a change that could free capital tied up in bridges and fast transfers, on the same day Solana switched on an upgrade targeting 150 milliseconds

Ethereum's researchers have put a faster settlement clock into the chain's formal upgrade pipeline. On October 21, 2026 Ben Edgington, who coordinates Ethereum's fast-finality research, opened a pull request in the Ethereum Improvement Proposals repository titled "Add EIP: Decoupled Consensus", with Francesco D'Amato, Roberto Saltini, Yann Vonlanthen and Mikhail Kalinin as co-authors, according to CoinDesk and the repository. The draft says its goals are to prepare for a post-quantum transition, improve fork-choice security and reduce time to finality, and that many details are still to be determined.

CoinDesk reports the initial target is a fourfold improvement, from about 16 minutes today to roughly four. Today each validator casts one vote that both helps choose the next block and helps lock in older ones, and a block is final only after two full rounds. The proposal separates those jobs into distinct votes, so the locking-in step can run on its own, faster schedule that developers can tighten over time.

Why it matters is money, not just speed. Services that pay out instantly on cross-chain transfers have to front capital until the source transaction is final. CoinDesk gave a worked example, noting that Ethereum Foundation research counts idle capital as a real cost: a service moving $1 million an hour has about $267,000 waiting on repayment with a 16-minute delay, but about $67,000 at four minutes. Circle's fast transfer route from Ethereum charges 0.01%, about $1 per $10,000 before network and application charges, and caps the value of transfers awaiting settlement, the outlet noted.

Four minutes would still trail Solana, whose Alpenglow upgrade went live on Wednesday targeting finality of about 150 milliseconds, CoinDesk said. What to watch next: editor review of the draft, on which a reviewer has proposed the number 8453, and whether a future hard fork adopts the design, which has no activation date.