Quantinuum’s Helios system has achieved 99.9995% logical memory fidelity, translating to just five uncorrectable errors per million operations and setting a new record for quantum computation, the company says. This result demonstrates a 12-fold reduction in logical memory errors with each tested increase in code distance, indicating a clear path toward scalable and reliable quantum computers.
“With our latest result, we are on the cusp of this monumental achievement,” says Quantinuum, believing they are nearing a “megaquop” machine capable of running over a million operations error-free and signaling the potential end of the Noisy Intermediate-Scale Quantum era.
Helios Achieves Record 99.9995% Logical Memory Fidelity
99.9995%, meaning only five uncorrectable errors occur per million operations, establishing a new world record for quantum computation and surpassing previous benchmarks set by superconducting systems by a factor of over 100. This level of accuracy, demonstrated by Quantinuum’s trapped-ion system, represents a substantial reduction in the rate of logical errors, the errors that persist even after error correction is applied, and brings the prospect of reliable, scalable quantum computers closer to reality, according to the company. The fidelity was measured during syndrome-extraction cycles, a critical step in identifying and correcting errors within the quantum memory.
Each tested increase in code distance reduced logical memory errors by an average factor of 12, and logical gate errors by an average factor of 9, demonstrating the exponential error suppression needed to scale toward longer, more reliable quantum computations. A member of the Quantinuum team said, “We felt a mix of excitement and suspense every time the results came in.
When errors are this rare, each one really matters.” These results indicate that an Apollo-scale machine could reach the fault-tolerant quantum advantage regime using 30 times fewer physical qubits than other leading approaches, reinforcing their path to Apollo, their first universal fault-tolerant quantum computer.
We felt a mix of excitement and suspense every time the results came in. When errors are this rare, each one really matters.
Distance Scaling Demonstrates Exponential Error Suppression
The rate, denoted as Λ, quantifies the exponential reduction in error rate with each increase in code distance; a single increase yields a 12-fold reduction, while a threefold increase results in a 1728-fold improvement. These results also surpass a prior color code demonstration, achieving a 2,200-fold improvement with a distance 5 code. These comparisons, conducted at equivalent code distances, demonstrate the superior performance of Quantinuum’s approach and its logical error-correction architecture.
Quantinuum’s Helios Hardware Exceeds Fault Tolerance Threshold
This result, built upon the company’s logical error-correction architecture, surpasses previous superconducting system records by a factor of over 100 and demonstrates a clear trajectory toward scalable, reliable computation. The team’s success stems from a flexible approach, showing multiple quantum error correction codes on the same commercial Helios hardware to optimize both accuracy and qubit efficiency, the company says. The ability to rapidly improve accuracy is critical as Quantinuum advances toward its 2029 goal of launching Apollo, its first universal fault-tolerant quantum computer.
The latest demonstration follows a Helix-code demonstration on Helios, highlighting the adaptability of their logical architecture and the precision of their trapped-ion technology. Projections based on these results suggest that an Apollo-scale machine could achieve fault-tolerant quantum advantage using approximately 30 times fewer physical qubits than competing approaches.
Quantinuum has also focused on optimizing the speed of decoding syndrome data, the process of interpreting error signals and applying corrections. The team successfully demonstrated accurate decoding with sub-millisecond latency, a speed 10 to 100 times faster than required, using machine-learning decoders developed by their in-house AI team, Quantinuum reports. This rapid response time is essential for preventing quantum information from degrading before corrections can be applied, and positions Helios as a “pathfinder” system, informing the development of future hardware generations.
Color Code Experiments Advance Toward Apollo Architecture
This level of accuracy was demonstrated through successive tests using color codes of increasing distance, 3, 5, and 7, with each increase in code distance yielding an average 12-fold reduction in logical failure rates. The system’s ability to consistently lower error rates with each iteration highlights a clear, exponential trajectory toward scalable and reliable quantum computation. A key metric driving this progress is Λ, a value representing the rate at which error rates decrease with increasing code distance; Helios demonstrated a Λ of 12, significantly surpassing the 1.56 achieved in a comparable superconducting architecture.
This disparity allows for reaching an error rate, such as 10⁻¹², considered viable for practical applications, with substantially shorter code distances, and therefore fewer physical qubits. Calculations show that to achieve an error rate of 10⁻¹², a superconducting system would require a code distance of 109, while Helios needs only a distance of 21.
The implications for qubit requirements are substantial; generating 100 logical qubits would necessitate approximately 900,000 physical qubits in the superconducting system, compared to roughly 30,000 for Quantinuum’s Quantum Charge Coupled Device (QCCD) processors, a 30-fold resource savings, by the company’s account. While the color code technically requires odd code distances, the team notes that reaching the “teraquop” regime, defined as exceeding one million error-free operations, necessitates a code distance of 21, the next available increment beyond the tested range.
These experiments are not isolated; Quantinuum has previously demonstrated multiple quantum error correction codes, all the necessary building blocks for logical computation, and has consistently lowered logical error rates with each generation of hardware, the company claims. The convergence of these advancements positions the company to move closer to its Apollo architecture, slated for completion in 2029, and potentially signals the end of the Noisy Intermediate-Scale Quantum (NISQ) era.
Trapped-Ion Technology Enables Robust Quantum Memory
This result surpasses the previous superconducting system record by more than 100-fold, demonstrating a substantial leap in the stability of quantum information storage. This rapid improvement stems from the system’s flexible logical architecture and the inherent accuracy of trapped-ion technology, as demonstrated following a Helix-code demonstration on the same Helios hardware, Quantinuum says. The previous color-code error rate at a distance of 5 was 0.819% per cycle, a figure now significantly surpassed by Helios’s performance.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
