Solana has migrated its developer network to Alpenglow, a major consensus upgrade designed to sharply reduce transaction finality times and change how validators coordinate across the blockchain. The migration was completed on September 25, with the activation — known as Alpenswitch — taking effect at epoch 1167 following a 17-minute countdown. The transition was carried out by engineers at Anza, the firm that develops core software for the Solana network, and Solana co-founder Anatoly Yakovenko highlighted the milestone in a brief social media post shortly after the migration.

Alpenglow is designed to cut Solana's transaction finality from roughly 12.8 seconds to between 100 and 150 milliseconds, potentially enabling near-instant confirmation for applications that depend on rapid settlement. Finality is the point at which a transaction is considered permanently settled and cannot be reversed, which makes it one of the core benchmarks for how responsive a blockchain feels in practice; moving that guarantee from several seconds to a fraction of a second would bring confirmation closer to the immediacy users expect from conventional internet applications.

The upgrade is now being tested on Solana's devnet, after the protocol had already been deployed on testnet. Running the new consensus system across both public testing environments gives developers and network operators the opportunity to examine its performance and stability before any decision is made about deploying it on the mainnet-beta network, the production environment where transactions carry real value.

Votor Replaces TowerBFT

The core of the change replaces Solana's longstanding TowerBFT consensus mechanism, which has secured the network since its mainnet launch in 2020, with Votor, a new voting system developed by Anza engineers in collaboration with researchers from ETH Zurich. Validators approved the change under Solana Improvement Document SIMD-0326, the network's formal mechanism for ratifying protocol changes.

Votor alters how validators exchange voting information. Rather than recording individual technical votes as transactions inside blocks, validators communicate those votes directly, and the information can then be consolidated into cryptographic BLS certificates — aggregated signatures that condense many individual votes into a single compact proof.

The change is intended to reduce the share of block capacity consumed by consensus-related activity. Technical voting transactions previously occupied a substantial portion of Solana's block space, with some estimates placing the figure as high as 75%. By moving this activity outside the transaction flow and aggregating it into compact certificates, Alpenglow is designed to free additional network capacity for user transactions while potentially lowering operational costs for validators.

Notably, the upgrade does not require a fundamental redesign of Solana's virtual machine or smart-contract architecture. Applications built on the existing Solana Virtual Machine can continue operating while the underlying consensus mechanism changes.

Sub-Second Finality Targets

Alpenglow's principal performance objective is a substantial reduction in the time it takes for transactions to become final. Under the new system, finality is expected in about 100 milliseconds when validators representing at least 80% of the network's stake respond within the required window. A fallback mechanism is designed to deliver finality in approximately 150 milliseconds when the higher threshold is not immediately met.

The shift to sub-second finality could prove particularly relevant for payment applications, trading systems, and Web3 games, where delays in transaction confirmation can directly affect the user experience.

The protocol also incorporates a 20+20 fault-tolerance model. Under this structure, the network is designed to remain stable if 20% of validators become unavailable while another 20% simultaneously face attacks or other disruptions.

Developers Begin Real-World Testing

The devnet deployment gives application developers a low-risk environment for testing software against the new consensus rules. Devnet tokens do not carry real monetary value, allowing teams to examine transaction behavior without exposing users to the financial risks associated with mainnet activity. Payment applications could use the environment to assess faster confirmation, while blockchain games could test whether lower settlement latency improves interactive experiences.

Not every application will require changes. Basic software that sends transactions or reads account balances can continue to operate without major modifications. However, analytics platforms and services that count blockchain transactions may need to adjust their methods, because validator votes will no longer appear as individual transactions within blocks.

Alpenglow is live on devnet >
— Solana (@solana) September 25, 2026

Data services will also need to account for competing candidate blocks until the network reaches final agreement; combining information from competing blocks prematurely could create inaccurate historical records.

Mainnet Timing Remains Uncertain

Alpenglow is now running on Solana's devnet and testnet, but no firm date has been announced for mainnet-beta deployment. Stability testing remains ongoing, and the cited performance figures are targets based on protocol testing and simulations rather than confirmed results under full live-market conditions. Actual user experience could also depend on wallet processing, exchange deposit procedures, and other infrastructure outside the consensus layer.

The public testing phase gives Solana developers and infrastructure providers an opportunity to validate whether Alpenglow's targeted speed and resilience can translate into reliable performance before the protocol is considered for mainnet deployment. The migration therefore represents a significant technical test for Solana rather than a completed network-wide upgrade, and the results from devnet and testnet will provide important evidence on how the new consensus architecture performs under broader workloads — and whether the network can move toward its targeted sub-second finality on live infrastructure. The markers to watch from here are whether devnet and testnet hold the targeted finality figures as testing workloads grow, and whether a subsequent governance step puts a mainnet-beta activation on a concrete schedule.