Solana activated Alpenglow's Votor consensus system on October 21, 2026, replacing TowerBFT and changing how validators reach agreement on transaction history. The Solana Foundation's subsequent engineering changelog lists Alpenglow among the features activated on mainnet. The rollout aims to shorten the wait for irreversible settlement, with a target of roughly 150 milliseconds under normal network conditions.
That is a consensus target, not a promise that every wallet, exchange or application will show a completed payment within 150 milliseconds. The distinction matters because an application must still receive the result and apply its own processing rules.
The activation had an epoch boundary and a separate handoff
The Foundation's October upgrade documentation places the start of epoch 1064 at approximately 13:36 UTC on October 21. An epoch is a recurring period of network operation. Activating the feature at that boundary did not immediately replace the old consensus system.
Validators continued using TowerBFT for the first 5,000 slots, about 18 minutes. Slot 459,652,999 became the certified genesis block for Alpenglow, and the new consensus took over from slot 459,653,000. The Foundation says the handoff itself took about 900 milliseconds.
The timing is therefore more precise than a single advertised launch hour: the feature was enabled first, followed by the consensus transition. These identifiers also distinguish the actual mainnet change from earlier development and test deployments.
Votor is running; Rotor remains a separate project
Alpenglow's design contains several components, but the October activation did not deliver all of them. The consensus proposal, SIMD-0326, explicitly covers Votor while excluding Rotor, a planned replacement for Solana's Turbine block-distribution system. Rotor is intended for a later rollout and does not yet have a scheduled release in the Foundation's update.
Votor changes voting and finalization. Validators exchange votes directly rather than putting them into on-chain transactions, and combine signatures into certificates that establish how much stake supported a decision. The design can finalize a block after one voting round with support representing 80% of stake, or use a second round with 60% thresholds.
Those thresholds refer to the voting weight of stake, not an equal head count of machines. The background to that distinction is explained in the guide to proof of work and proof of stake. Alpenglow replaces the previous Proof-of-History and TowerBFT consensus arrangement; it does not replace staking with mining.
Faster finality reaches applications through their infrastructure
Anza's commitment-status documentation now treats the network's confirmed and finalized states as equivalent under Alpenglow. Both identify a transaction committed to the permanent ledger. Processed remains a different state: it indicates execution, without that same finality assurance.
For applications that previously waited for finality, the change can reduce the gap between seeing a transaction and treating its result as settled. Ordinary users do not need to migrate wallets or tokens. The Foundation says transaction execution, programs and fee mechanics remain unchanged. The broader Solana profile explains SOL's role in fees and staking.
Infrastructure still affects what a user observes. Anza documents different ways for nodes outside the admitted validator set to learn that a block is final. A node relying on certificates carried in later blocks can learn the outcome a block interval later, or longer if blocks are skipped. Direct connections to admitted validators can reduce that delay, but network latency remains.
Consequently, protocol finality, an infrastructure provider's notification time and a customer's completed checkout are different measurements. A meaningful performance comparison must identify which one was timed.
Validator costs shift to an admission fee
Removing vote transactions changes the operator's payment mechanism. Anza's current vote-account guide specifies a Validator Admission Ticket costing 0.8 SOL per epoch, burned from the validator's vote account. That balance must be available above the account's rent-exempt minimum.
The charge in one epoch pays for admission in the next. An underfunded validator cannot vote or produce blocks in that following epoch. Anza also warns that commission income may not cover the fee, so routing revenue into the account does not eliminate the need to monitor it.
The effect on a validator's total operating costs cannot be inferred from the admission fee alone. Hardware, connectivity and service arrangements remain separate expenses; the relevant comparison is the complete operation before and after migration.
What the first days establish, and what they do not
As checked on October 31, Solana's official status page lists the mainnet cluster as operational and shows no reported incidents from October 21 through October 31. That is a useful record of the initial period, but it is not a measurement of every application's latency or a guarantee against future disruption.
Anza describes roughly 150-millisecond finality as a target under normal network conditions. Sustained measurements across network conditions, validator locations and infrastructure providers are still needed to assess the experience users actually receive. The concrete change is already in place: validators are reaching agreement through Votor. Rotor and the longer operating record remain ahead.