Solana moves Alpenglow finality upgrade onto public testnet

Solana has activated its Alpenglow consensus upgrade on the public testnet, beginning a broader test of a system designed to reduce transaction finality from about 12.8 seconds to a target of 150 milliseconds. The Solana Foundation’s upgrade page lists the testnet activation as live and mainnet activation as not yet active.

Finality is the point at which the network’s rules treat a transaction as irreversible. Exchanges can wait for it before crediting deposits, and bridges before releasing assets on another blockchain. The shorter target could let those services act sooner, if the upgrade performs as designed.

How will Alpenglow change the way validators agree?

Alpenglow replaces TowerBFT, Solana’s current consensus system, with a voting protocol called Votor. In the existing system, validators record votes as onchain transactions, and enough votes must accumulate across 32 slots before a block becomes final. Solana’s documentation says that process takes about 12.8 seconds.

Under Votor, validators send votes directly to one another and combine them into certificates. A block can then become final after one or two rounds of voting. In plain terms, validators confirm their agreement among themselves instead of waiting for votes to build up in the blockchain’s record. The finality target is about 150 milliseconds; that measures how quickly a transaction becomes final, not how often Solana produces blocks.

The upgrade changes consensus, not transaction execution. The Solana Foundation says the SVM, transactions, programs and fees remain unchanged. Applications and wallets therefore do not need to change how they send transactions to use the network.

What is the testnet meant to prove?

The public testnet lets developers and operators try the migration across a larger environment of validators and services. According to CoinDesk’s report on the test, Alpenglow had already run for more than four months on a smaller community cluster created for the upgrade. The public test checks whether the broader network can switch together.

Validators joining the initial test need Agave 4.3, the main Solana validator software maintained by Anza. CoinDesk reported that the alternative validator clients Firedancer and Frankendancer do not yet support this Alpenglow test. The first migration will therefore rely on Agave, rather than testing multiple client implementations at once.

When could Alpenglow reach Solana’s main network?

There is no confirmed mainnet launch date in the sources. CoinDesk reported that Sept. 28 was a tentative date for activating Agave 4.3 features on mainnet, not a confirmed date for Alpenglow itself. The Foundation’s upgrade page likewise lists mainnet as not activated. Until that changes, 150 milliseconds remains a test target, not the finality time users should expect on Solana’s main network.