BSC Rolls Out Pasteur Hard Fork; QANet Tests Show 88% Higher Throughput With 450ms Blocks Unchanged

BNB Smart Chain has activated the Pasteur hard fork on mainnet, bundling three BNB Evolution Proposals aimed at tightening bridge verification and validator authorization while expanding effective block capacity without altering the network's 450-millisecond block time. At the center of the upgrade is BEP675, which revises the block submission workflow between specialist builders and validators. Under the prior flow, builders executed transactions before proposing a block, and validators then re-executed the same transactions before signing, consuming time within the fixed block window. Pasteur introduces an alternative path where validators first check consensus rules and proceed with full execution verification after signing, reducing duplicated work. Builders can still use the legacy route. BNB Chain says the change is designed to help blocks carry more transactions during peak demand without increasing the 100 million gas limit or extending block intervals. In QANet testing—BNB Chain's internal environment intended to mirror geographically distributed validators—throughput increased about 88%, rising to 2,324 transactions per second from 1,237. Average gas used per block climbed from 46.35 million to 84.15 million, while both the block interval and the gas ceiling remained unchanged. The team emphasized these figures reflect controlled tests rather than mainnet performance. Pasteur also includes two security-focused proposals. BEP682 blocks duplicate validator entries during cross-chain light block verification, preventing the same validator from being counted more than once in bridge approvals. BEP695 strengthens controls around validator key rotation, slashing, and governance voting, including removing authority from old validator keys and preventing restricted addresses from voting. The release continues BSC's recent performance-oriented cadence. The Maxwell upgrade reduced average block time from 1.5 seconds to roughly 0.8 seconds in June 2025, and the later Fermi upgrade lowered it further to 450 milliseconds. Pasteur keeps that timing intact and instead targets fuller blocks and tighter authorization. Real-world impact will depend on how quickly builders and validators adopt the new submission path under sustained mainnet demand, and whether higher throughput can be delivered under congestion without compromising consensus discipline.