The limitations of silicon single-photon detectors are now sharply defined; tens of nanoseconds of recovery time restrict path-based quantum random number generator designs to tens of megabits per second. This bottleneck becomes stark when contrasted with macroscopic photodetector designs achieving 80 Gbps, a raw data rate demonstrating a substantial speed advantage. Ma et al. caution that output tests can never stop a malicious generator from emitting a string chosen in advance. Jacak et al.’s paper lists attacks tied to weak random numbers, including the 2015 theft of 18,866 bitcoins from the Bitstamp exchange.
Quantum Random Number Generator Designs and Performance
Macroscopic photodetector designs achieve raw data rates of 80 Gbps, a speed significantly exceeding the 180 Mbps attained by single-photon detector arrival-time QRNGs. Ma et al. sort practical designs by detector type. Jacak et al. argue these devices are pseudorandom as well, because classical physics is deterministic.
Glossary entry describes running a Bell test on a quantum computer to prove randomness quality, and says Quantum Origin product uses this method. This contrasts with the speed-focused designs detailed by Ma et al., which often require a trade-off between data rate and the level of trust placed in the device’s inherent randomness.
A comparison of five designs revealed a substantial performance gap; the fastest design outputs bits approximately 170 billion times faster than the self-testing Bell test method. However, these figures represent lab conditions and do not necessarily translate directly to real-time applications. Quip’s QVRF service, a randomness oracle for smart contracts, employs a four-step pipeline to deliver quantum-derived random values, beginning with circuits run on quantum hardware, according to the company.
The service then hashes samples into a Merkle tree, publishes the root, and performs cross-entropy benchmarking on re-simulated samples using classical hardware. Finally, a Toeplitz extractor condenses the audited samples into near-uniform bits, a process designed to ensure the integrity and unpredictability of the generated randomness. According to quip.network/qvrf, the service’s documentation details the steps, emphasizing the importance of rigorous auditing and extraction techniques.
The speed of randomness extraction remains a critical challenge for many designs; Ma et al. note that none of the phase-noise designs achieve real-time performance due to limitations in the extraction process. This highlights a fundamental tension in QRNG development: maximizing data rate while maintaining a high degree of statistical confidence in the generated randomness. The choice of design, therefore, depends on the specific application and the acceptable balance between speed and trust.
Quip’s QVRF Pipeline and Trust Limitations
The service begins by running circuits on quantum hardware to generate raw samples, then hashes these samples into a Merkle tree, publishing the root for transparency. Following this, a random subset of samples undergoes re-simulation on classical hardware and is evaluated using cross-entropy benchmarking, a process designed to assess the quality of the generated randomness. Despite the multi-layered approach, Quip itself functions as both the oracle and the trustee, creating a single point of trust for users, according to quip.network/qvrf.
This centralization contrasts with the ideal of fully decentralized randomness, where multiple independent parties verify the process. The verification of a contract transcript remains a work in progress, with checks currently performed after delivery rather than in real-time at the moment of value exchange. This post-delivery verification means potential manipulation, while detectable, occurs before confirmation, a critical distinction for high-security applications.
The design, rooted in a certified randomness protocol published in Nature in 2025 and demonstrated on a 56-qubit trapped-ion processor, acknowledges its own limitations in its documentation. Quip publishes its commitment on-chain before delivering the random values, a measure intended to prevent value swapping without detection. But the documentation also states that certification parameters, the request interface, and the fee model are still being finalized.
This ongoing development suggests the service is still evolving, and users should be aware of potential changes. The service does not yet meet the cryptographic definition of a verifiable random function, instead being described as a physical randomness source with a verification record. The team at Quip invites potential integrators to review the QVRF documentation and request access at quip.network/qvrf to assess the trust model before implementation.
Jacak et al. argue that even hardware-based generators can be considered pseudorandom if classical physics is deterministic, highlighting the ongoing challenge of achieving truly unpredictable randomness. While Quip’s approach uses quantum mechanics, the current architecture necessitates careful consideration of the trust assumptions placed in the service operator.




See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
