The upgrade also lifts testnet throughput from 1,237 to 2,324 transactions per second and tightens the rules on how validator keys are retired.
BNB Smart Chain activated its Pasteur hard fork at 02:30 UTC on Tuesday, and the headline change is not a speed boost. It is a repair to how the chain counts validator signatures on cross-chain messages, a gap that until this morning let the same validator vote twice on a bridge attestation.
Node operators had to be on client version v1.7.7 before the fork block to keep syncing. The upgrade bundles three BEPs, and the one attracting most scrutiny is BEP-682. It rejects duplicate validator entries during cross-chain light-block verification, the process BSC uses to confirm what happened on the Beacon Chain before executing a bridge withdrawal. Under the old logic a single validator's signature could be submitted more than once and pass as separate approvals, thinning the security guarantee that a quorum of independent signers is supposed to provide. Anyone who has followed bridge exploits over the past three years knows how that class of bug ends.
BEP-695 tackles the other end of the validator lifecycle. When a validator rotates its signing key, the old key's permissions are now removed on the same block; any pending slashing penalties transfer to the new key; and addresses under restriction cannot vote in governance from the retired key. Before Pasteur, an operator caught misbehaving could in principle rotate to a fresh key, shed accountability and continue voting on protocol changes. That path is closed.
The performance work sits in BEP-675, which allows block builders to submit pre-executed blocks. Validators verify rather than re-execute, which fills the block more completely within the same gas limit and slot time. Testnet numbers put the improvement at 1,237 to 2,324 transactions per second, roughly a doubling, without any change to fees or block cadence for users. Real mainnet throughput will depend on how many builders actually adopt the new payload format, but the ceiling has moved.
BNB Chain has framed Pasteur as a step toward its stated goal of doubling mainnet capacity this year. The chain has spent much of 2026 chasing capacity upgrades against a backdrop of security incidents on adjacent products; the BNB Chain tutorial wallet that bankrolled a $628,000 memecoin trade in early August was an embarrassment the team could have done without, and cross-chain bridges remain the most fragile surface on any smart-contract network. Fixing the duplicate-signature path before an attacker found it is the correct sequence.
Bridge exploits have been the single largest category of crypto losses since 2022. The Ronin bridge attack on Axie Infinity alone drained $625 million, and the Wormhole exploit on Solana added another $320 million. Most bridge failures do not exploit cryptography; they exploit signature accounting, key custody or lax verification of the messages that connect one chain to another. BEP-682 is a narrow patch to exactly that kind of accounting error. It does not make BSC's bridges secure by itself, but it removes one class of defect from the base layer.
For users, the fork is invisible. No token migration, no address changes, no new asset. Wallets, exchanges and dApps already on v1.7.7 continued to sync without interruption. Any node still on an older client forked off at block activation and stopped seeing new blocks, and needs to reinstall and resync from a snapshot to rejoin. That is the standard cost of missing a hard fork window, and BNB Chain's operator communications had been running for weeks.
What Pasteur does not address is the harder governance question around BSC's validator set, which remains small enough that observers have long argued the security model relies more on Binance's operational integrity than on the mathematical assumptions the code makes. Twenty-one active validators is not a large surface. Tightening the rules on what those validators can do is useful; changing the number of them is a different debate, and one BNB Chain has not opened.
The next scheduled BNB Chain upgrade is due later this quarter and is expected to target further throughput gains rather than security primitives. Pasteur, for now, is the fix that shipped.
