Quantinuum Demonstrates Fault-Tolerant Quantum Computing on Commercial System

London, UK – September 13, 2026

Quantinuum has experimentally validated its Helix quantum error correction architecture on the Helios quantum computer, achieving record-breaking fidelities that demonstrate the first complete demonstration of logical memory, logical computation, and logical entanglement on a commercial quantum processor without relying on post-selection techniques.

The validation, announced this week by the quantum computing company, marks a critical transition from theoretical error correction codes to practical, scalable fault-tolerant quantum computing. Using its 98-physical-qubit trapped-ion Helios system, Quantinuum's team demonstrated that the Helix architecture can preserve encoded quantum information at error rates lower than the underlying physical operations—the defining requirement for fault tolerance.

The Challenge of Quantum Error Correction

Quantum computers are extraordinarily sensitive to environmental disturbances. Even tiny fluctuations in temperature, electromagnetic fields, or cosmic rays can flip qubit states and corrupt calculations. To address this, researchers developed quantum error correction (QEC), which encodes a single logical qubit across multiple physical qubits and continuously monitors for errors.

The challenge has been that earlier QEC demonstrations required either post-selection (discarding runs that showed errors) or produced error rates higher than the physical qubits themselves. Neither approach scales to practical quantum computing. The Helix architecture addresses both problems through a heterogenous design.

"With this demonstration, we have put all the pieces together," said a Quantinuum representative, pointing to a foundation for scalable, fault-tolerant quantum computing.

A Heterogenous Code Architecture

Traditional quantum error correction codes typically apply the same encoding scheme to all computational operations. The Helix architecture diverges from this approach by using different codes optimized for different tasks. The core Helix code handles standard Clifford gates efficiently, while a separate code prepares the "magic" states required for universal quantum computation.

This division of labor significantly reduces the physical qubits and time overhead. Researchers constructed the three-logical-qubit GHZ state—a benchmark for multi-qubit entanglement—spanning both the surface code and the Helix code. Chain-mapped gates linked these disparate encodings, yielding a logical GHZ fidelity of 99.925% with an upper bound of 99.975%.

The Helix code itself is constructed by concatenating stabilizer codes in a specific combination designed to optimize performance. Its core capability is minimized spacetime volume, achieved through exotic entanglement schemes and reconfigurable qubit connectivity. Unlike traditional QEC codes relying on simple, two-dimensional entanglement networks, Helix utilizes more complex configurations comparable to a cat's cradle, reducing physical qubit requirements per logical qubit.

Record Logical Memory Performance

Beyond logical operations, the Helios system demonstrated unprecedented logical memory performance. Through 20 rounds of syndrome extraction—the process of monitoring for errors without collapsing the quantum state—the team recorded a per-qubit, per-round error rate of 4.6 × 10⁻⁵ without post-selection.

Further refinement, employing a modest 0.5% post-selection, reduced the block logical error per round to 1.9 × 10⁻⁵, highlighting potential for even greater fidelity with minimal data loss. These results establish that logical information can be preserved longer than the physical qubits that encode it—the essential prerequisite for meaningful quantum algorithms.

Efficient Logical Gates Through Transversal Operations

The Helix architecture accelerates logical computation using transversal gates and automorphisms. Transversal gates perform operations across logical qubits without requiring complex physical interactions, while automorphisms accomplish operations through software-level qubit relabeling. These techniques enable logical circuits to run with fewer physical resources and in less time than many alternative approaches.

Experimentally, researchers benchmarked the complete logical Clifford group—all gates excluding T gates—while interleaving up to 27 rounds of syndrome extraction. Both gate operations and qubit idling contribute to errors, so reductions in these areas directly translate to lower logical error rates. A substantial portion of logical circuits was executed using only physical single-qubit gates and qubit relabeling, effectively realizing some operations "for free" through ion transport and software adjustments.

This approach drastically reduces both the qubits needed for encoding and the overall circuit complexity, lessening demands on physical resources and shortening runtime.

From Helios to Apollo

The validation was conducted on the Helios system, currently available to Quantinuum customers through cloud services and on-premises offerings. The team expects the same code to be even more performant when implemented on the forthcoming Apollo system, which will build on the current results achieved with Helios.

The company highlights that this demonstration represents a harmonious whole encompassing a candidate architectural code, logical computation capabilities, multiple encodings within a single architecture, and compatibility with commercial hardware. The architecture's design allows for logical computation without requiring a corresponding increase in the number of encoded logical qubits, a significant advantage for complex algorithms.

Beyond Physical Qubit Limitations

The demonstration extends beyond simply achieving logical operations. Researchers successfully preserved encoded quantum information for extended periods, mitigating a primary source of error on the Helios platform: leakage. Through circuit-level reduction units and new leakage repump capacity, the team further underscored the practical viability of the Helix architecture.

Quantum memory performance remains a fundamental building block for advancing fault-tolerant quantum computing. Beyond memory, the Helix architecture delivered a logical error rate that represents a quantifiable improvement over the physical two-qubit Clifford error rate on Helios, again achieved without post-selection.

This efficiency gain is partially attributable to the team's adaptive syndrome extraction technique, which reduces the number of physical gates required for each logical gate. The reduction in gate count also shortens the physical runtime, enabling more complex algorithms to run within the coherence time of the underlying hardware.

A Path Toward Practical Quantum Computing

Importantly, these results were not obtained on a specialized testbed or limited-functionality hardware. The experiments were conducted on the same Helios system currently available to Quantinuum's customers for their research. This means the demonstrated capabilities are not theoretical projections but present-day capabilities that developers can access today.

Simulations suggest that anticipated improvements in physical fidelity with the upcoming Apollo system will further align logical error rates with the company's long-term roadmap targets. The foundation laid by Helix is intended to support this next-generation system, suggesting a clear path toward increasingly powerful and reliable quantum computers.

The low logical error rates observed, coupled with practical logical operations, validate the company's roadmap and suggest a clear path toward building increasingly powerful and reliable quantum systems. This marks a transition away from the limitations of the Noisy Intermediate-Scale Quantum (NISQ) era, where quantum processors produced results that required classical post-processing to extract meaning.

Industry Implications

The validation carries direct implications for enterprises developing quantum algorithms. By demonstrating that logical operations can outperform physical operations without post-selection, the team has shown that practical quantum advantage is achievable with current hardware. This shifts focus from theoretical feasibility to implementation challenges.

Companies building quantum software can now target logical operations rather than physical qubits, enabling them to design algorithms that will work across different quantum architectures. The ability to dynamically adapt to the demands of different quantum algorithms and error profiles through reconfigurable connectivity represents a significant step toward commercial quantum computing.

Scientific Collaboration

The Helix architecture was developed by Quantinuum's research team, building on earlier work in trapped-ion quantum computing. The demonstration involves scientists from Quantinuum's engineering and science teams, with validation performed on production hardware. This integration of research and commercialization represents a maturation of the quantum computing industry.

Quantinuum highlights that the Helix code's design allows for logical computation without requiring a corresponding increase in encoded logical qubits, a significant advantage for complex algorithms. The architecture's reconfigurable connectivity allows the system to dynamically adapt to the demands of different quantum algorithms and error profiles.

The Road Ahead

As Quantinuum frames its development roadmap, it relies on the ability to combine efficient logical gates with multiple QEC encodings. The validation of Helix on the Helios system represents a critical step toward realizing practical, scalable fault-tolerant quantum computing. The team's success in implementing efficient gate types and demonstrating logical computation represents a step toward realizing the full potential of fault-tolerant quantum computing, moving beyond isolated results and toward a functional, scalable architecture.

The company believes this validation of a fault-tolerant architecture on real hardware signifies a substantial move toward quantum computers capable of tackling meaningful problems at scale. With the foundation established by the Helix demonstration, the quantum computing industry can now focus on scaling logical qubit counts and expanding the range of algorithms that can benefit from error-corrected quantum computation.

Source: Quantum Zeitgeist, September 13, 2026. Quantinuum, "Introducing Helios: The Most Accurate Quantum Computer in the World," November 5, 2025. Nature, "A 98-qubit trapped-ion quantum computer with all-to-all connectivity," 2026.