Psy Protocol Claims 521,000 TPS in Public Stress Test, Opens Data for Verification

Psy Protocol says it has reached 521,000 transactions per second (TPS) during a large-scale network stress test powered by Google Cloud, positioning the result as one of the highest publicly verifiable throughput benchmarks reported for a Proof-of-Work-based blockchain architecture.

The Hong Kong-based project announced that the test simulated hundreds of thousands of concurrent users using globally distributed Google Cloud infrastructure. According to Psy, the results are fully reproducible, with raw proof data publicly available for independent verification.

While many blockchain networks have published peak TPS figures in recent years, analysts have frequently noted the difficulty of comparing performance claims across projects. Theoretical maximums, short-burst laboratory tests, sustained throughput targets and real-world averages are often presented under different conditions and infrastructure setups.

Recent announcements from networks such as MegaETH and Solana have highlighted this dynamic. MegaETH has promoted high headline capacity figures alongside separate sustained performance tests conducted under defined parameters. Solana, for its part, publicly distinguishes between theoretical ceilings, observed real-world averages and occasional peak stress results.

Such distinctions are not uncommon in performance benchmarking, but context can fade once a headline number circulates. Psy is attempting to differentiate its claim by publishing the composite zero-knowledge proofs underlying the stress test, allowing external observers to independently examine how the 521,000 TPS benchmark was derived.

From re-execution to proof verification

Traditional blockchains require each node to re-execute transactions to confirm their validity. Psy’s architecture takes a different approach.

Instead of re-executing transactions across the network, Psy uses what it calls a PARTH state model combined with client-side proving. Transactions are proven locally on a user’s device using zero-knowledge cryptography, and those proofs are then recursively aggregated on-chain.

Because verifying a succinct ZK proof is computationally cheaper than executing each transaction individually, Psy argues that its scaling curve grows logarithmically rather than linearly as transaction volume increases.

In practical terms, this means higher throughput without proportionally increasing network workload.

Designed for the “Agentic Internet”

Psy situates the milestone within what it describes as the “Agentic Internet,” a scenario in which autonomous AI agents execute high-frequency microtransactions, negotiate payments and coordinate economic activity directly on-chain.

In this context, the 521,000 TPS result is presented as evidence that the network’s architecture can sustain transaction volumes associated with automated, programmatic activity. Rather than focusing solely on a headline throughput figure, the test is framed as validation of a scaling model designed to handle high concurrency without proportionally increasing verification overhead.

Verifiable, not just theoretical

One of the more unusual aspects of Psy’s announcement is its focus on reproducibility rather than just raw throughput. The team says any composite proof generated during the stress test can be independently verified using consumer-grade hardware, without relying on private auditors or internal reports.

Instead of asking observers to trust a benchmark figure, Psy has published the proof data and methodology behind the test. According to the project, anyone can inspect the composite proof outputs and validate the performance claims directly through its public explorer.

The project has also referenced a $100,000 bounty tied to disproving the benchmark, positioning it as an additional incentive for independent scrutiny of the results.

The test was conducted on Psy’s public testnet using Google Cloud infrastructure, and the network has not yet launched mainnet. While the benchmark was achieved in a public test environment, Psy’s decision to release the underlying proof data introduces a higher standard of transparency around performance claims. In a market where peak, theoretical and sustained throughput figures are often difficult to compare, independently verifiable benchmarks may play a growing role in how scalability is assessed across blockchain networks.