Sui’s 40.6M TPS Test Was Followed by Slightly Lower SUI Closes

Sui (SUI) recorded peak throughput of 40,614,180 transactions per second during an Oct. 7 live stress test in Singapore, while available historical data for Binance’s SUI/USDT pair listed a slightly lower close on Oct. 8.

The data listed a close of $1.1282 on Oct. 7 and $1.1271 on Oct. 8. The page does not specify the daily candle’s time zone, so the two readings show price movement around the demonstration without establishing a causal relationship.

The 40.6 million TPS figure came from programmable Sui Tunnels rather than 40.6 million independent transactions processed directly on Sui’s base layer. The tunnels handle activity off-chain and settle their opening and closing states on Sui mainnet.

More than 10,000 tunnels opened on Sui mainnet during the demonstration. The design is intended to support applications and artificial-intelligence agents that may need to conduct frequent interactions before settling activity on-chain.

Adeniyi Abiodun, chief product officer and co-founder of Mysten Labs, said, “No market in the world needs 6 million transactions per second, let alone 40 million: people just don't move that fast. Agents do.”

The test exceeded a previous peak of 6,086,766 TPS recorded on July 4. Both measurements were taken inside Sui Tunnels, so they do not represent conventional base-layer transactions executed independently by Sui validators.

The network also supports gasless peer-to-peer transfers for allowlisted stablecoins. Those transfers allow users to move the assets without holding SUI to pay transaction fees, limiting the direct connection between some network activity and immediate token use.

Sui said CertiK independently reviewed the test in real time and would examine the proofs, execution logs, journals and other supporting evidence before publishing a report. The 40,614,180 TPS result remains subject to that pending examination.

The test followed plans by Mysten Labs and Google Cloud for a verifiable-agent system designed to document whether artificial-intelligence agents act within authorized limits. The planned system was described in an earlier report on the verifiable-agent project.