Woofun AI reports that Sui achieved a throughput milestone of 6.08 million TPS during a public live stream on July 4, utilizing its programmable tunnel technology to demonstrate extreme scalability.

The official figure of 6,086,766 TPS represents a six-fold increase over the project’s initial target of one million TPS. This performance marks a substantial leap from a previous benchmark of 297,000 TPS, which was recorded in a more isolated environment. The new record is approximately 20 times higher than that earlier test, underscoring the network’s capacity for high-volume applications.

Structurally, the test facilitated millions of agent-to-agent interactions settled on-chain through the programmable tunnel. The Sui team emphasized that this architecture allows AI agents to operate independently, eliminating the need for human input.

This shift highlights the potential for automated, autonomous transaction processing within machine-to-machine economies.

Per Woofun AI, such capabilities are transformative for decentralized finance (DeFi) and supply chain management, sectors that rely on rapid, trustless interactions. The demonstration confirms that Sui’s layer 1 networks can support the high throughput required for mass-market decentralized applications. This positions the protocol as a contender in building infrastructure for complex, automated ecosystems.

In comparative terms, Visa’s network processes around 24,000 TPS at peak, while Solana has claimed theoretical peaks above 50,000 TPS. Sui’s 6 million TPS result is orders of magnitude higher than these established benchmarks.

However, the programmable tunnel appears to be a specialized infrastructure component designed for specific high-throughput use cases, rather than a blanket improvement across all network activities.

While the live-streamed nature adds transparency often missing in benchmark claims, it remains to be seen how this performance translates under sustained, decentralized, real-world conditions. Practical deployment and network decentralization will be key factors in determining long-term impact. For now, the achievement serves as a proof of concept for blockchains handling millions of autonomous agent interactions per second.