Solana Alpenglow reached mainnet on Wednesday, October 21, at the start of epoch 1064. The upgrade replaces TowerBFT, the voting system Solana has used since launch. Its target is roughly 150 milliseconds to finality. That figure compares with about 12.8 seconds under the old design, an improvement of roughly 85 times. Galaxy calls it the largest change to Solana’s core protocol since launch. The numbers matter, but the mechanics behind them matter more.
Alpenglow is now live on Solana mainnet. ⛰️ Finality just went from ~12.8 seconds to ~150ms. The biggest consensus upgrade in Solana’s history, built by @anza_xyz.
— Solana (@solana) October 21, 2026
What Activated on Mainnet
Solana’s upgrade page lays out a staged handoff. TowerBFT keeps control for about 5,000 slots after the epoch begins, roughly 18 minutes. Alpenglow then takes over at slot 459,653,000. Published start times for the epoch vary between sources, so operators should check the official schedule. Validators needed Agave 4.3.2 or a compatible fork to join the switch.
This first phase ships only Votor, the voting component. Rotor, the planned block propagation redesign, has no date yet. Meanwhile, Solana separately cut block time from 400ms to 200ms, a change that finished in October. Execution stays the same, so the Solana Virtual Machine, transaction formats, programs, and fees do not change. Importantly, the 150ms number is a design target. Real mainnet measurements will show how close the network gets.
How Votor Finalizes Blocks
Under TowerBFT, validators cast votes as ordinary transactions. According to Galaxy’s analysis, those votes made up about three quarters of network transaction volume. Votor removes them from blocks. Validators now exchange BLS-signed votes directly and combine them into compact certificates. The next block carries the certificate instead of thousands of vote transactions.
Two finalization paths run at the same time. If 80% of stake votes to notarize a block, it finalizes in a single round. If participation falls short, the network needs a 60% notarization certificate and a separate 60% finalization certificate. Whichever path completes first wins. As a result, finality depends mostly on network latency rather than on slot timing.
Anza engineer Brennan Watt describes the logic in plain terms. In a Solana Compass interview, he said validators “actually wait for these certifications before considering a fork valid.” He added that this “drastically simplifies the fork selection.” Consequently, forking should rarely appear. Community test cluster data, reported by Solana Compass, showed 96% of blocks taking the fast path. That figure came from a community validator citing a dashboard, so readers should treat it as indicative.
The Safety Trade Behind the Speed
Alpenglow changes what the network tolerates. TowerBFT stayed safe with up to 33% adversarial stake, but it offered no extra cushion for offline validators. Alpenglow tolerates up to 20% adversarial stake and a further 20% offline stake. Combined, the network can lose 40% of stake to faults and keep running.
Designers made this choice deliberately. Solana’s past outages came mostly from bugs, congestion, and correlated crashes, not coordinated attacks. However, a lower hostile-stake threshold is a real concession, and critics can reasonably debate it. Additionally, slashing for conflicting votes is defined but stays switched off in this first release. Anza says its research partners, including a team from ETH Zurich, aimed for a design that is provably correct.
What Changes for Validators
The economics shift as well. Validators used to pay a fee on every vote, which Galaxy estimates at about 1 SOL per day at 400ms slots. Alpenglow replaces that with the Validator Admission Ticket from SIMD-0357. The ticket costs 0.8 SOL per epoch, and the network burns it from the vote account.
Operators must keep at least 0.8 SOL above the rent-exempt minimum in the vote account, not the identity account. A validator that misses the payment cannot vote or produce blocks for the following epoch. Notably, the consensus set caps at 2,000 validators, while mainnet currently runs about 700. Firedancer’s 26.10 release supports Alpenglow, although the migration itself ran on Agave. In contrast, Frankendancer loses support once Alpenglow is active.
What Builders and Users Will Notice
Ordinary users need to do nothing. Signing, sending, and approving transactions work the same way. Developers will see more change. Vote transactions leave blocks, so reported transaction counts and TPS will drop without any real fall in activity. Teams that compare historical charts should adjust for that.
Additionally, the confirmed and finalized commitment levels now converge. Solana’s guidance tells indexers to buffer streamed data by slot and bank ID, and to keep the version that reaches confirmed. Proof of History ticks also leave the ledger, and the clock timestamp now comes from the leader. Developers can check migration status with the getAgGenesisCert call, which returns null while TowerBFT still runs.
What Comes Next
Phase two aims to speed leader handover, so the next leader can start building optimistically. Phase three brings Rotor, which flattens block propagation into fewer network hops. Slashing remains the main open safety item. Meanwhile, the first weeks of live data will decide whether 150ms holds up under real load and global validator distribution. For payments and trading applications, that measurement matters more than any target.
*Disclaimer: News content provided by Genfinity is intended solely for informational purposes. While we strive to deliver accurate and up-to-date information, we do not offer financial or legal advice of any kind. Readers are encouraged to conduct their own research and consult with qualified professionals before making any financial or legal decisions. Genfinity disclaims any responsibility for actions taken based on the information presented in our articles. Our commitment is to share knowledge, foster discussion, and contribute to a better understanding of the topics covered in our articles. We advise our readers to exercise caution and diligence when seeking information or making decisions based on the content we provide.
