Quantinuum has recorded a logical memory fidelity of 99.9995% on its Helios quantum computer, marking a major step in the effort to make quantum information more reliable. The measurement means Helios experienced about five logical failures for every million error-checking cycles.
That number gives the result a clear meaning. A logical memory is designed to protect one logical qubit by using many physical qubits, and the Helios experiment tested how well that protection held up through repeated correction rounds. Quantinuum measured the fidelity at 99.9995% per error-checking cycle, setting a new company record.
The result also cuts the company’s previous reported record by a factor of ten. In other words, the new error rate is one-tenth of the earlier figure, showing that the improvement is not a small adjustment to an existing result.
How the Helios experiment measured reliability
The largest memory experiment used 73 physical qubits to protect one logical qubit. Quantinuum’s estimate came from three failures across 29,000 experimental shots, with each shot containing 20 correction rounds. Those details matter because the reported figure rests on repeated trials rather than a single test.
From that experiment, the central error estimate reached five failures per million cycles. The one-standard-deviation interval extended approximately from three to nine failures per million. That range shows how the estimate changes when the uncertainty in the measurement is taken into account, while the central result remains five per million.
The experiment therefore combines a high fidelity figure with a clear account of the observed failures. Helios did not simply produce a headline percentage; the test also recorded the number of shots, the number of correction rounds, and the failures used to calculate the estimate.
For quantum computing, that kind of measurement is central. Quantinuum is not only measuring whether a logical qubit can hold information, but also tracking how the memory behaves as the system performs error checks. The Helios result places the focus on the relationship between physical qubits, logical protection, and repeated correction.
Why the physical layout matters
Helios uses trapped ions with connectivity that allows interactions beyond a fixed nearest-neighbor layout. That feature forms part of the setting in which the memory experiment took place, alongside Quantinuum’s control of the physical qubits.
David Hayes described the result in direct terms: “Helios-1’s record-breaking performance partly follows from the Quantinuum team improving its control over the physical qubits.” His comment points to the hardware control behind the reported fidelity, rather than treating the result as a gain from the error-correction method alone.
Hayes also said, “It’s super exciting.” Steven Flammia offered a similar assessment, saying, “This is definitely an impressive technical advance.” Together, the comments reflect the importance of the result without changing what the experiment measured: 99.9995% logical memory fidelity per error-checking cycle.
Quantinuum has also demonstrated that logical qubits can have decreasing error rates as more physical qubits are added. The company recorded a twelve-fold decrease when increasing the number of physical qubits from seven to 19. That finding connects the Helios result to a wider pattern in the company’s work: adding physical resources can improve the protection of logical information.
What comes next for Quantinuum
Quantinuum’s next quantum computer, Sol, will be nearly twice as large as Helios-1. The planned increase gives the company a larger system in which to pursue the same basic goal: protecting logical qubits with physical qubits while keeping error rates under control.
The Helios result does not remove the need for more testing. Its one-standard-deviation interval stretches from approximately three to nine failures per million, so the central estimate is only one part of the measurement. Still, the record provides a precise benchmark for future experiments involving logical memory and error correction.
The main message is simple. Helios stored one logical qubit using as many as 73 physical qubits, ran correction rounds across thousands of shots, and reached a measured fidelity of 99.9995%. With Sol set to be nearly twice as large as Helios-1, Quantinuum now has a new benchmark to carry forward.
Based on
