BNB Smart Chain is heading into its most security-focused upgrade of the year. The Pasteur hard fork is scheduled to activate at 02:30 UTC on August 25, 2026, and every node operator on the network needs to be running client version 1.7.7 before that moment or risk falling out of consensus entirely.
Packaged under the umbrella proposal BEP-673, the fork bundles three separate BNB Evolution Proposals into a single coordinated cutover: BEP-682 for cross-chain bridge security, BEP-695 for staking and governance hardening, and BEP-675 for transaction throughput.
Closing a Bridge Validator Loophole
BEP-682 targets a specific weakness in the signature verification used by BSC's cross-chain bridge, where a manipulated validator set could list the same signer multiple times and clear the approval threshold with far fewer genuine validators than the protocol assumes. Pasteur closes that gap by rejecting duplicate validator entries outright — a fix that lands at a pointed moment, just three days after The Sandbox's own cross-chain bridge was exploited on Base and BNB Smart Chain, underscoring how bridge logic remains one of the most attacked surfaces in crypto.
Staking, Governance and a Throughput Jump
BEP-695 hardens staking and governance after key rotations: validators that retire a consensus key lose its administrative authority immediately, pending penalties stay attached through a rotation instead of being shed, and blacklisted addresses are barred from joining signature-based governance votes. BEP-675 is the performance piece, letting block builders submit pre-executed blocks for validator verification ahead of full finalization. In testnet benchmarks, that cut the time validators spent on the immediate check from 125 milliseconds to 15 milliseconds and lifted throughput from 1,237 to 2,324 transactions per second, while block time held steady at 450 milliseconds.
Related: Sandbox Exploit Mints $49B in Fake SAND, But Actual Loss Is Tiny
Pasteur is the latest in a string of 2026 upgrades aimed at squeezing more performance out of BSC. The Fermi hard fork in January cut block times to 0.45 seconds, and the Osaka/Mendel upgrade in April stabilized performance at that faster cadence — together pushing benchmark throughput to roughly 5,200 TPS in the first half of the year, with BNB Chain developers stating the goal is to double mainnet throughput again in the second half.
The fork isn't landing untested. BSC's Chapel testnet activated an earlier version of Pasteur on July 21 running client v1.7.6, giving developers roughly five weeks of live data before Monday's mainnet cutover — a buffer the team is counting on given how quickly bridge exploits elsewhere have escalated this year.
