Solana moves Alpenglow finality upgrade into testnet trials

Solana has begun moving its Alpenglow consensus upgrade into public testnet trials, testing a design intended to cut finality from about 12.8 seconds to roughly 150 milliseconds. CoinDesk reported the testnet rollout on September 23, while noting the network’s tracker still listed the switch as pending at the time. The target is a protocol estimate, not a measured mainnet result.

Finality is the point when a block can no longer be reversed under the protocol. Exchanges can use it to decide when to credit deposits, and bridges when to release funds on another chain. Solana’s current TowerBFT system reaches that stronger guarantee after votes accumulate across 32 slots, taking about 12.8 seconds. Faster “confirmed” status arrives earlier, but carries a weaker guarantee.

How would Alpenglow shorten the wait?

Alpenglow replaces TowerBFT’s onchain vote transactions with Votor, a system in which validators send votes directly to one another and combine them into certificates. The Solana Foundation’s upgrade guide says Votor can finalize a block in one or two voting rounds. If at least 80% of stake votes to notarize in the first round, the fast path can finalize it; otherwise, a second round uses 60% stake thresholds.

The change is confined to consensus: the Foundation says transaction execution and the Solana Virtual Machine remain unchanged. The 150-millisecond figure is an expected finality time based on the stake distribution used in testing, and can vary with network conditions. It is not a new block-time promise. The guide describes Rotor, a planned replacement for Solana’s block-propagation system, as a later phase.

What does the first test leave out?

Validators joining the initial migration need Agave 4.3, the main Solana validator software maintained by Anza. CoinDesk reported that the test does not yet support Firedancer or Frankendancer, alternative validator clients developed by Jump Crypto. That makes the first migration narrower: it tests the upgrade across Solana’s public test network, but through Agave rather than across different client implementations.

Client diversity matters because separate implementations can reduce the chance that one software fault affects the whole network. Alpenglow also changes what infrastructure reads: validator votes will leave blocks and travel through certificates instead. The Foundation says execution is unaffected, but systems that inspect block contents or vote transactions will need to account for the new format.

What would show the upgrade is ready?

The next signals are whether the public testnet migration completes cleanly, how reliably Votor finalizes blocks under live network conditions, and when Firedancer and Frankendancer add support. CoinDesk said September 28 was a tentative date for a separate Agave 4.3 mainnet feature update, not a confirmed Alpenglow launch date. Mainnet timing and the 150-millisecond target therefore remain unproven.