The biggest throughput figure a blockchain security firm has ever signed off on is sitting on CertiK’s homepage right now: 40.68 million transactions per second.
It isn’t CertiK’s own number, and it isn’t an on-chain number either. If you read the 2,000-word report behind it, you’ll come to become more interested in what CertiK chose not to verify than in what it did.
A Record Number Hiding in Plain Sight
Scroll past the client logos and media mentions on CertiK and you’ll find it in the “Latest Research And Market Insights” strip. One card says CertiK independently verified Sui’s Mainnet Tunnel performance campaign, which recorded more than 15.2 billion off-chain transactions across 7,680 tunnels at a peak of 40.68 million TPS.
Most readers will see “CertiK” next to “40.68 million TPS” and stop there. Crypto social media mostly did the same. Sui’s own post described the figure as set live on stage at Sui Basecamp 2026 and certified by CertiK, and added that the network now clears in under 20 seconds what Visa processes in a day.
Walking through CertiK’s verification method when the report first came out. Now that the number has been promoted to CertiK’s front page, it’s worth looking at exactly what it represents.
What Sui Actually Ran on October 7
According to CertiK’s announcement, Sui ran a live Mainnet performance campaign on October 7, 2026, during Sui Basecamp Day. It recorded 15,241,348,297 tunnel transactions across 7,680 tunnels, and every tunnel was opened and settled on chain.
A Sui Tunnel works roughly like a private room. A small group of participants carries out many application actions without sending each one to the blockchain. Sui records the starting terms when the tunnel opens and the final result when it closes. Everything in between happens off-chain as signed receipts. CertiK compares the design to Bitcoin’s Lightning Network but notes that tunnels can handle more than payments.
The test was more varied than a single payment loop. The campaign ran 10 runs across five workload types: simple payments, flash, blackjack, tic-tac-toe and battleship. Payments still made up most of the activity. Its 6,144 tunnels committed more than 12 billion transactions, or 78.98% of the total.
How CertiK Got From Raw Data to 40,682,328
This is the part of the report I respect most. The peak came from the busiest complete second, between 08:46:17 and 08:46:18 UTC, when 40,682,328 qualifying tunnel transactions were committed. The highest 60-second rolling average was 40,339,143.65 TPS.
Doing the math on the gap between those two figures. It’s about 343,000 transactions, or roughly 0.84%. In other words, the network held near-peak throughput for a full minute rather than spiking for one second.
CertiK also avoided the easiest way to inflate a number like this. For concurrent tunnels sharing the same time window, it added up their transaction counts once and divided by a single shared duration. It did not add individual per-tunnel rates together or multiply one tunnel’s rate by the number of tunnels.
The checking itself was thorough. All 7,680 executions and all 960 slots reached VERIFIED status, and the 15.24 billion committed transactions matched the full-history admitted receipts exactly. The work ran on dedicated AWS infrastructure with 128 vCPUs and 247 GiB of RAM.
There’s also a small correction in the story. Sui’s first announcement reported 40,614,180 TPS. CertiK’s audited peak came in about 68,000 higher, a 0.17% difference. A verifier landing slightly above the project’s own claim is unusual, and I think it reflects well on the method.
The Fine Print the Homepage Leaves Out
The homepage card doesn’t mention any of this, but the full report is clear about its limits.
First, this isn’t on-chain throughput. CertiK states that Tunnel TPS measures off-chain tunnel transactions and is not a measurement of Sui’s on-chain capacity.
Second, the data came from the project being tested. The Sui team supplied the campaign data and executed all test scenarios, and CertiK examined the retained evidence and recalculated the totals.
Third, the cryptographic checks were not exhaustive. CertiK reconciled the arithmetic, receipt counts and timestamps, but it did not re-run individual signatures or do a fresh cryptographic replay of every receipt.
Fourth, the report can’t speak to how the system would behave in normal use. The records don’t establish clock accuracy, failure rates, latency percentiles or results across repeated runs. The sustained figure is the best 60-second window observed, not a guaranteed minimum.
Put simply, CertiK verified that the accounting is correct. It did not verify that Sui can handle 40 million real users’ actions every second.
The Ratio That Tells the Real Story
The number that’s most telling isn’t the TPS figure. More than 15.2 billion off-chain transactions were anchored to Sui through just 1,920 on-chain lifecycle transactions: 960 openings and 960 settlements.
By simple calculation, that’s about 7.9 million off-chain actions for every on-chain transaction. That ratio is what the tunnel design is built to achieve. It’s an impressive engineering result, but it measures how well Sui compresses activity, not how fast its base layer runs.
Why This Number Isn’t Comparable to CertiK’s Past Records
This isn’t the first speed record CertiK has signed off on, and the earlier ones show why comparisons are tricky. In its earlier TON verification, CertiK worked from per-minute transaction count data and arrived at 104,715 TPS for TON’s testnet. In April 2025, Qubic announced 15.52 million TPS verified by CertiK, measured over 10 peak ticks on its Layer 1 mainnet.
On paper, Sui’s tunnel figure is about 2.6 times Qubic’s and nearly 390 times TON’s. But these were three different kinds of tests: a sharded testnet, a base-layer burst and an off-chain channel network. They don’t belong on the same leaderboard.
My view is that the 40.68 million figure is real, well documented and carefully bounded. It’s also a measure of what an off-chain system can process, not what Sui’s blockchain itself handles. CertiK’s report is honest about that, even if the one-line homepage card isn’t.
Disclosure: This is not trading or investment advice. Always do your research before buying any cryptocurrency or investing in any services. Follow us on X @nulltxnews
Be the first to comment