Solana has advanced its Alpenglow consensus upgrade to the threshold of public testnet deployment, giving developers and validators a broader environment to test a major shift in transaction finality. The upgrade is designed to cut finality time from roughly 13 seconds to about 150 milliseconds.
What Alpenglow Changes
Solana currently relies on TowerBFT, a consensus mechanism that requires validators to record votes on-chain. Under TowerBFT, validators must accumulate enough votes across 32 slots before a block is considered final. Alpenglow replaces this with a new protocol called Votor, which lets validators exchange votes directly rather than depending on the same on-chain voting sequence.
With Votor, agreement can be reached in one or two voting rounds. One round can finalize a block when sufficient validator stake participates; a second round offers an alternative path when participation is lower. Anza estimates median finality near 150 milliseconds, with earlier simulations reaching about 100 milliseconds under favorable conditions.
From an application or wallet perspective, the upgrade is designed to be mostly invisible. Transactions continue through the same interfaces; the notable changes take place in validator communication and consensus.
Expanding Testing to Public Testnet
Alpenglow has been running for more than four months on a smaller community cluster built specifically for consensus testing. Moving to Solana’s public testnet brings more validators and infrastructure providers into the evaluation process. Because testnet tokens hold no real-world value, developers can restart the network, inspect failures, and test migration procedures without risking mainnet assets.
Anza first moved Alpenglow into community validator testing in May, describing it as the largest consensus change in Solana’s history. Validators participating in the test are expected to run Agave 4.3, the latest branch of Anza’s validator software.
Agave 4.3 and Mainnet Timeline
Anza recommended Agave 4.3 for broader mainnet validator adoption on September 21. The rollout initially targeted operators representing 10% of mainnet stake, and the recommendation later expanded to 25%. By August, developers expected the 150-millisecond finality improvement to arrive through Agave 4.3.
Anza’s schedule lists September 28 as a tentative date for resuming mainnet feature activation, though the date is subject to change. That date should not be read as confirmation that Alpenglow will activate on Solana mainnet. As of early Wednesday, Anza’s feature gate tracker still listed Alpenglow testnet activation as pending.
Finality is a separate measure from other performance upgrades. Solana recently reduced its target slot time from 300 milliseconds to 250 milliseconds via SIMD-0525, bringing the network to four slots per second. That change also shortened each validator’s four-slot leader window from 1.2 seconds to one second. SIMD-0525 ultimately targets 200-millisecond slots, or five slots per second, but no confirmed mainnet activation date has been announced for that stage.
Alpenglow follows a separate process through SIMD-0326. Rather than shortening slots, it replaces TowerBFT with Votor and changes how validators agree that blocks have reached finality.