The Pasteur hard fork on BSC activated at 02:30 UTC on 25 August 2026, patching a bridge verification flaw that could have let a crafted validator set pass approval thresholds with fewer real signers than required, and introducing a block-building route that nearly doubled throughput in controlled testing. BNB Chain’s official blog confirmed the activation and noted that node operators needed to be running BSC client version v1.7.7 beforehand or risk dropping off the network.
What the Pasteur Hard Fork BSC Bundle Actually Fixes
Three BEP proposals, grouped under the BEP-673 upgrade plan, make up Pasteur’s scope.
BEP-682 targets the cross-chain verification precompile at address 0x67. That precompile counts signatures against a validator set to approve bridge transactions. Before Pasteur, it did not check for duplicate entries, meaning a crafted set could list the same validator repeatedly and clear the approval threshold with fewer distinct signers than required. The fix simply rejects any validator set where a single entry appears more than once.
BEP-695 tightens validator key rotation. According to cryptonews.net, the three specific changes are: a rotated old validator key now loses its admin privileges immediately rather than retaining them; restricted addresses are blocked from governance voting; and slashing controls on key rotation are hardened. Taken together, the two security BEPs close the class of attacks where stale authority or duplicate accounting could be weaponised at the bridge or consensus layer.
BEP-675 is the capacity workstream. It introduces a separate block-submission path for specialist builders, and sits inside BNB Chain’s stated H2 2026 goal of doubling mainnet throughput, with a longer-term 10x improvement targeted across the broader BNB Chain ecosystem.
How BEP-675 Restructures Block Production
Under the previous route, a builder executed all transactions, submitted a proposed block, and then the validator executed them again before signing. BNB Chain’s team said that repeated execution consumed a large share of the 450-millisecond block window, frequently leaving blocks underfilled during congested periods.
BEP-675 lets builders submit pre-executed blocks. The validator checks the proposed block against consensus rules, signs and broadcasts it, then completes full execution verification afterwards. Builders retain the option to use the old path, so the change is additive rather than a forced cutover.
In testing on QANet, BNB Chain’s internal environment designed to mirror the geographically distributed validator set, the new route lifted throughput from 1,237 to 2,324 transactions per second, an increase of roughly 88%. Average gas used per block rose from 46.35 million to 84.15 million. The block interval and 100-million gas ceiling were unchanged. BNB Chain cautioned that these figures came from a controlled workload and do not represent live mainnet performance.
The upgrade had been running on BSC’s Chapel testnet since 21 July 2026, where client v1.7.6 was used for the testnet activation, giving operators roughly five weeks of pre-mainnet exposure before the v1.7.7 mainnet requirement landed. The full announcement, including the August 25 date and version requirement, was published in a BNB Chain blog post on 14 August 2026.
Pasteur Hard Fork BSC in the Context of BSC’s Block-Time Compression Sequence
Pasteur is the latest step in a multi-fork push to compress BSC’s block cadence and expand capacity. The Maxwell hard fork moved block times from 1.5 seconds to 0.75 seconds, the snippet’s reference to “about 0.8 seconds” diverges from BNB Chain’s own Maxwell announcement, which sets the target at 0.75 seconds. The Fermi hard fork then activated on 14 January 2026 at block height 75140593, cutting block times further from 0.75 seconds to 0.45 seconds (450 milliseconds) and tightening fast finality rules.
Pasteur leaves the 450-millisecond block time intact. The gain here is not a faster clock but a fuller block, exploiting headroom created by the earlier latency reductions without requiring another consensus-layer overhaul.
BSC client release notes on GitHub and the crypto.news recap both confirm the mainnet activation. The next test of BEP-675’s design will come during BSC’s next high-congestion period: whether pre-executed block submission holds up under adversarial conditions in production is a different question from what QANet showed.
