TL;DR
- BNB Smart Chain activated Pasteur, combining three proposals that strengthen bridge verification, validator authorization and block-building efficiency without changing its 450-millisecond block time.
- BEP-675 lets builders submit already-executed blocks, while validators check consensus rules before completing full execution verification, reducing duplicated work.
- QANet tests showed throughput rising about 88% from 1,237 to 2,324 transactions per second, though BNB Chain stressed that these were controlled tests rather than mainnet performance measurements.
BNB Smart Chain has activated its Pasteur hard fork, combining three BNB Evolution Proposals aimed at closing bridge verification gaps, tightening validator authorization and increasing block capacity without changing the network’s 450-millisecond block time. BEP-682 rejects duplicate validator entries during cross-chain light-block verification, while BEP-695 strengthens controls around key rotation, slashing and governance voting. The upgrade’s core message is unusually broad: BSC is trying to improve security and throughput at the same time. Old validator keys lose authority, restricted addresses cannot vote, and duplicate validators can no longer be counted twice in bridge approvals.
Pasteur hardfork is now live on BSC Mainnet.
The upgrade strengthens bridge, staking and governance security, while giving blocks more room to carry transactions at the same 450ms block time.
Read the full upgrades
https://t.co/41QWQh1GGLpic.twitter.com/mTTnRb9PNk
— BNB Chain (@BNBCHAIN) August 25, 2026
Pasteur targets fuller blocks without slowing the network
Pasteur’s performance push comes through BEP-675, which changes how specialist builders submit blocks to validators. Previously, builders executed transactions before proposing a block, then validators executed them again before signing, consuming valuable time inside the block window. The new route removes that duplication before signing, allowing validators to check consensus rules first and complete full execution verification afterward. Builders may still use the previous pathway, but the alternative is designed to help blocks carry more transactions during busy periods without raising the 100-million gas limit or extending block intervals during the network’s busiest periods.

Tests on QANet, BSC’s internal environment designed to mirror geographically distributed validators, showed the new route lifting throughput by about 88%, from 1,237 to 2,324 transactions per second. Average gas used per block rose from 46.35 million to 84.15 million while the block interval and gas ceiling stayed unchanged. Those numbers make the capacity gain look substantial, but they are controlled-test results rather than mainnet measurements. The distinction matters because real network conditions can differ, leaving Pasteur’s production performance to be demonstrated as builders and validators adopt the new submission path under sustained transaction demand on mainnet.
The hard fork also extends BSC’s recent sequence of performance-focused upgrades. Maxwell reduced average block time from 1.5 seconds to about 0.8 seconds in June 2025, while the later Fermi upgrade brought it down to 450 milliseconds. Pasteur keeps that timing intact while targeting fuller blocks and stronger authorization controls. Instead of chasing another block-time reduction, BSC is now trying to extract more useful capacity from the same interval. For operators and users, the practical test will be whether the verification safeguards and builder changes deliver higher throughput under real congestion without weakening consensus discipline.

