BNB Chain activated its Pasteur hard fork on BNB Smart Chain (BSC) mainnet at 02:30 UTC on August 25, 2026. The public Chapel testnet got it first, on July 21, 2026. Node operators had to run BSC client v1.7.7 to stay on the canonical chain.
Pasteur bundles three changes: a fix to how BSC checks cross-chain bridge proofs, tighter rules for validator key rotation and governance, and a new way for block builders to hand validators pre-executed blocks. The headline number, an 88% jump in throughput, comes from the third change and from internal tests.
Before and after Pasteur
The table sets the reported test figures beside the rule changes. The numbers are test results from QANet, an internal network that both Cointelegraph and Yellow say mirrors BSC’s validator geography. Yellow’s lead calls the figure “testnet throughput,” while its body ties it to QANet, so we avoid calling it a public testnet result.
| Item | Before | After | Source |
|---|---|---|---|
| Throughput (QANet test) | 1,237 TPS | 2,324 TPS | Cointelegraph |
| Avg gas per block (test) | 46.35 million | 84.15 million | Cointelegraph |
| Validator share of critical path | 125 ms | 15 ms | Yellow |
| Bridge proof check | Could count a validator more than once | Rejects duplicate validator identities | Cointelegraph, InfStones |
| Block interval and gas limit | 450 ms; 100 million | Unchanged | Cointelegraph |
The three changes
Bridge fix (BEP-682). BEP-682 changes the precompiled contract at address 0x67, which validates CometBFT light blocks used for cross-chain verification. It now rejects inputs that contain duplicate validator identities. Cointelegraph describes the purpose this way: one validator can no longer be counted more than once in bridge approvals. InfStones, a node operator, adds that the check covers validator address, consensus public key, a non-zero BLS public key, and a non-zero relayer address. None of the coverage we opened reports that this flaw was exploited. The fix closes a verification gap, and it doesn’t tell us anyone used it.
Validator governance (BEP-695). BEP-695 hardens BSC’s staking and governance contracts. Per InfStones, old consensus keys lose their validator administration rights after rotation, slashed validators are evicted after a key rotation, and the governance blacklist check applies to the actual voter in signature-based voting.
Faster block handoff (BEP-675). BEP-675 adds a SendBidBlock path. Builders can submit blocks they have already executed, so validators can check, sign, and broadcast without re-running every transaction first. Cointelegraph notes the older route remains available. InfStones says it is enabled per builder through RPC and is not itself a consensus hard fork.
Who Benefits and How
Most of the benefit goes to people who run or depend on the plumbing: validators and node operators, builders, and teams that use the cross-chain bridge. The coverage addresses them, not ordinary BNB holders, and it describes no action for holders. The bridge and governance changes are live on mainnet since August 25, 2026. The throughput gain is a test result only, so it’s not something users can count on yet. We found no sourced claim about transaction fees changing because of Pasteur.
Where it falls short
The 88% figure is a test result. Cointelegraph reports throughput rose from 1,237 to 2,324 transactions per second in QANet tests, and Yellow quotes the team’s caution that these came from “a controlled test workload rather than live mainnet traffic.” Average gas used per block rose from 46.35 million to 84.15 million in the same tests, while block interval and gas limit stayed put.
The 125 ms to 15 ms figure is narrow too. Yellow reports it as the validator’s share of the critical path measured on QANet after builders began handing over executed blocks. It isn’t end-to-end latency. Mainnet gains are unmeasured: InfStones calls a roughly 50% gas limit ceiling a modeled estimate, not a guaranteed mainnet result.
Quick answers
Do BNB holders need to do anything?
The coverage we opened addresses validators, node operators, and bridge teams, not ordinary holders. Node operators had to upgrade to v1.7.7 before activation.
Was any money stolen through the bridge bug?
The sources don’t report an exploit. They describe the change as a hardening of verification.
Will BSC really handle 2,324 TPS now?
Not shown. That number comes from controlled internal tests. Real traffic varies, and no mainnet measurement was in the sources.
Which version and date matter?
Client v1.7.7, activated on BSC mainnet at 02:30 UTC on August 25, 2026, after a July 21, 2026 run on the Chapel testnet.
Sources
- Cointelegraph: BNB Chain activates Pasteur hard fork to strengthen bridge security (Aug 25, 2026)
- Yellow: Pasteur Fork Ships To BNB Chain With A 2,324 Transactions Per Second Test Mark (Aug 25, 2026)
- InfStones: BNB Chain Pasteur Hard Fork, August 2026 (accessed Oct 13, 2026)
- Crypto Briefing: Pasteur hardfork scheduled for BSC mainnet at 02:30 UTC on August 25 (Aug 14, 2026)
