Pasteur Targets Bridge Security and Fuller Blocks
BNB Smart Chain is set to activate its Pasteur hard fork on Aug. 25, bringing a mix of bridge-security fixes, tighter validator controls and a new block-building mechanism designed to squeeze more transactions into each block without shortening block times again.
The upgrade is scheduled for 02:30 UTC on Aug. 25, 2026, according to BNB Chain. Mainnet node operators have been told to move to BSC client version 1.7.7 before activation. Pasteur has already been running on the Chapel testnet since July 21.
Unlike some of BNB Chain’s recent upgrades, Pasteur is not primarily about making blocks arrive faster. The Fermi hard fork in January cut BSC’s block interval from 0.75 seconds to roughly 0.45 seconds, while the subsequent Osaka/Mendel upgrade focused on stabilizing network behavior and improving finality around those sub-second block times.
Pasteur instead tries to make better use of that existing speed.
The upgrade groups three BSC-specific proposals under BEP-673. Two — BEP-682 and BEP-695 — require the hard fork itself. A third, BEP-675, introduces a new block-building path that becomes available after Pasteur and can then be enabled through RPC interfaces as builders integrate it.
BEP-682 addresses a weakness in the way BSC verifies validator sets when checking light blocks from compatible chains.
The existing verification system assumes that every validator in a submitted validator set is unique. But before Pasteur, the precompile performing that check did not explicitly enforce uniqueness. A specially constructed validator set could therefore include the same validator more than once, allowing its voting power to be counted repeatedly and potentially making a minority appear to have reached a required supermajority.
Pasteur closes that route by rejecting duplicate validator addresses, consensus keys, BLS keys and relayer addresses where applicable.
BEP-695 deals with a different set of edge cases involving validators, staking and governance.
After the upgrade, an old consensus key that has been replaced by a validator will no longer retain administrative authority. Slashing and validator eviction will also continue to apply properly after a consensus-key rotation, preventing a validator from avoiding removal simply by changing keys. The proposal additionally extends blacklist checks to governance votes submitted through signature-based methods.
The part users are more likely to feel indirectly is BEP-675.
Under BSC’s current block-building process, a specialist builder executes transactions while preparing a proposed block. The validator receiving that block then executes the transactions again before signing it. With only about 450 milliseconds between blocks, that duplicate work consumes part of the time available to assemble the next block.
BEP-675 changes the process so a builder can send an already executed block. The validator checks the proposed block against consensus requirements, signs and broadcasts it, while full verification is completed afterward.
In BNB Chain’s QANet testing environment, the change cut the validator’s portion of the critical processing path from 125 milliseconds to 15 milliseconds. Throughput increased from 1,237 transactions per second to 2,324 TPS, while average gas used per block rose from 46.35 million to 84.15 million under a 100 million gas limit. BNB Chain said finality lag did not increase.
Those figures remain testnet benchmarks, however, not observed mainnet performance.
For ordinary users and most application developers, BNB Chain says no migration is required. Block builders that want to use the new BEP-675 route will need to operate a full node rather than a fast node.
Node operators face the more immediate deadline. Besides installing version 1.7.7, operators upgrading older configurations need to remove the `[Eth] EnableBAL` setting from `config.toml`, which otherwise causes the client to fail at startup. Several older command-line options have also been removed or made ineffective.
Pasteur also fits into a broader performance push. BNB Chain said in its second-half 2026 roadmap that BSC had already reduced its block interval to 450 milliseconds and brought in-memory finality to around 650 milliseconds during the first half of the year. Its next objective is to double mainnet throughput, with a longer-term goal of delivering much larger capacity gains across the ecosystem.
One piece of background in older descriptions of BNB is now outdated. BNB Beacon Chain is no longer an active parallel chain. BNB Chain completed its Fusion process in 2024, migrating staking and governance functions to BSC and retiring the Beacon Chain.
BNB remains the native asset used for gas and staking on BSC and across the wider BNB Chain ecosystem.
Pasteur Is Really About Removing Waste — and Closing Small but Important Gaps
Pasteur does not have the obvious marketing appeal of cutting block times in half.
Users will not open a wallet on Aug. 25 and suddenly see transactions occurring twice as fast. In fact, BNB Chain is deliberately leaving the 450-millisecond block interval alone.
That may be exactly why the upgrade matters.
Once a blockchain has pushed latency this low, simply making blocks arrive faster becomes increasingly expensive. Validators are spread across regions, messages still have to travel over real networks, and every additional millisecond squeezed out of consensus leaves less room for things to go wrong.
Pasteur takes a different approach: remove work that does not need to happen twice.
If a builder has already executed a block, making the validator repeat the same work before it can sign consumes scarce time without creating additional block space. BEP-675 is essentially an attempt to reclaim that time and turn it into usable capacity.
The test results are eye-catching. A rise from 1,237 TPS to 2,324 TPS is close to an 88% improvement under the conditions BNB Chain tested.
But that number should not be treated as a promise that BSC’s real-world throughput will nearly double the moment Pasteur activates.
Mainnet traffic is messy. Transactions have different computational costs, builders behave differently, validator connections vary, and BEP-675 itself still depends on builders adopting the new path. The meaningful number will be how much more block capacity BSC actually uses during periods of sustained demand after the system has been running in production.
The security changes may ultimately be more important than the TPS headline.
BEP-682 fixes the kind of problem that can look almost trivial in code — checking whether the same validator appears twice — but becomes dangerous when that code sits underneath a bridge verification system responsible for deciding whether assets can move across chains.
BEP-695 follows the same pattern. An expired validator key keeping permissions, or a slashing action failing to follow a key rotation, sounds like an obscure operational problem until somebody deliberately uses it.
Pasteur therefore looks less like a radical redesign and more like protocol housekeeping at scale: eliminate duplicate execution, eliminate ambiguous validator identities and eliminate stale privileges.
That is also a natural progression for BSC. Earlier upgrades concentrated on getting the network from multi-second blocks to 450 milliseconds and sub-second finality. Once that foundation exists, the next question becomes whether the chain can use those milliseconds efficiently without weakening its security assumptions.
For BNB itself, the hard fork does not mechanically create a new source of demand or change the token’s economics overnight. Any longer-term benefit is more indirect. If BSC can process heavier activity while keeping transactions inexpensive and avoiding congestion, applications have more room to grow — and more on-chain activity means more use of BNB for gas and staking.
The first thing to watch after Aug. 25, then, is not the BNB price.
It is whether the upgrade activates cleanly, how quickly builders adopt BEP-675 and whether the large efficiency gains seen on QANet survive contact with real mainnet traffic. That will tell investors and developers far more about Pasteur than the headline benchmark alone.