Sui said its programmable offchain tunnels reached 40.6 million transactions per second during a live stress test designed to simulate high-frequency activity between AI agents, according to a Cointelegraph report published on October 7, 2026.
Disclaimer: News content is based on public news sources and provided for information only. It does not constitute investment advice.
Sui, a Layer 1 blockchain, announced that its programmable offchain tunnels reached 40.6 million transactions per second during a live stress test, according to Cointelegraph on October 7, 2026. The test was designed to simulate high-frequency activity between AI agents, a scenario in which software programs transact or exchange messages rapidly with minimal human input. The 40.6 million TPS figure is a benchmark result from that specific test environment. It refers to the tunnels rather than to Sui's base layer, which is an important distinction when comparing the result with ordinary onchain throughput. The announcement did not, in the summary provided, include independent verification or full details of the test setup.
Offchain tunnels are programmable execution channels that handle activity outside a blockchain's main ledger, potentially reducing congestion on the base network. In Sui's test, these tunnels were used to process a simulated stream of high-frequency interactions between AI agents rather than ordinary user transactions. This matters because performance measured offchain can be structurally different from performance measured on the Layer 1 itself, where consensus and data availability constraints apply. The summary does not specify whether the test involved real economic value, production hardware, or third-party validators, so the result should be read as a technical benchmark claim. Sui described the tunnels as programmable, suggesting they can be configured for different types of automated activity.
AI agents are software programs that can initiate actions, including payments or data exchanges, without direct human approval for each step. A network of such agents could produce bursts of machine-to-machine activity that look different from retail payment traffic, such as many small, frequent transactions. By simulating that pattern, Sui is testing its infrastructure against a use case that emphasizes throughput and low latency rather than occasional large transfers. The test does not establish that AI agents are already generating traffic at this scale on Sui; it is a controlled simulation. The result therefore speaks to a design target and a technical capability under test conditions.
Transactions-per-second benchmarks can vary widely depending on transaction complexity, batching methods, hardware, and whether the transactions are processed onchain or offchain. A 40.6 million TPS result from offchain tunnels is not directly comparable to the throughput of Sui's base Layer 1 or to other networks' mainnet performance. The absence of independent verification in the provided summary means the figure should be treated as a project-reported milestone. Readers can also note that stress tests are designed to push systems under specific conditions, not to mirror average real-world demand. Such announcements are common in blockchain development as teams seek to demonstrate scalability for future applications.
TPS stands for transactions per second, a common metric for measuring how many operations a system can process in one second. In this case, Sui reported that its programmable offchain tunnels reached 40.6 million TPS during a live stress test. The figure applies to that test environment and not necessarily to Sui's base Layer 1 under normal conditions.
No. Offchain tunnels are described as programmable channels that process activity outside the main ledger, while Sui's base Layer 1 handles onchain transactions under consensus rules. This distinction is important because offchain systems can avoid some base-layer constraints and may produce different performance results. The reported 40.6 million TPS therefore should not be read as Sui's mainnet throughput.
AI agents can generate machine-to-machine activity at speeds and frequencies that differ from human-driven transactions. A stress test simulating those agents is intended to explore whether the infrastructure can handle many small, rapid interactions. The test is a controlled simulation, so it does not mean AI agents are already producing that level of traffic on Sui.
Sui Reports 40.6M TPS in AI-Agent Stress Test
What Offchain Tunnels Do in This Context
Why AI Agents Are Used as a Stress-Test Scenario
How to Interpret the 40.6M TPS Figure
Sui's Broader Scalability Context
This article is rewritten from public news sources and further processed by AI. It is provided for information only and does not constitute investment advice — refer to the original source for the authoritative version.
Sui is a Layer 1 blockchain developed by Mysten Labs and uses the Move programming language along with an object-centric data model. Its design aims to process independent transactions in parallel, which is relevant to high-throughput use cases. The tunnel test adds to a broader industry effort to benchmark blockchains for automated, high-frequency activity. However, the reported 40.6 million TPS figure is specific to offchain tunnels in a stress test. It does not by itself indicate base-layer performance, production readiness, or real-world adoption. Based on public news sources.
Not necessarily. Stress tests are designed to push a system under specific conditions, and their results depend on transaction types, hardware, and whether activity is onchain or offchain. Real-world performance also depends on network conditions, adoption, and independent verification. The announcement reports a benchmark milestone rather than a guaranteed production throughput level.
Sui is a Layer 1 blockchain developed by Mysten Labs that uses the Move programming language and an object-centric data model. Its architecture is designed to support parallel processing of independent transactions. The recent tunnel test is part of ongoing efforts to demonstrate scalability for high-frequency use cases.
Readers should look at the test methodology, including whether the transactions were onchain or offchain, what hardware was used, and whether results were independently verified. TPS figures are difficult to compare across networks because transaction complexity and measurement standards can differ. Such announcements are best treated as technical claims about specific test conditions rather than universal performance guarantees.
