# Quantinuum Measures About Five Memory Errors Per Million Correction Cycles

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/34889
Published: 2026-10-19T16:32:29.000Z
Updated: 2026-10-19T16:32:29.000Z
Section: Technology

> Three failures across 580,000 correction rounds produced the rounded rate estimate; the test measured memory fidelity, not one million general-purpose quantum gates.

Quantinuum’s 98-qubit Helios processor reached an estimated logical-memory error rate of about five per million correction cycles, a result that measures progress in protecting quantum information.

The company announced the result at 8 a.m. ET (12:00 UTC) Monday, Oct. 19. The 98-physical-qubit trapped-ion system used a distance-7 color code. It measured 99.9995% logical memory fidelity, equal to an error probability of 5 × 10⁻⁶ per syndrome-extraction cycle. A logical qubit spreads information across multiple physical qubits, where repeated checks can expose and correct faults.

Researchers recorded three failures in 29,000 shots, each containing 20 correction rounds. That totals 580,000 round instances, yielding about 5.2 failures per million rounds, rounded to about five. The rate is an estimate from the sample, not five failures observed during a million-cycle test.

Across code distances 3, 5 and 7, the logical-memory error rate fell by an average factor of 12 at each tested increase. Greater code distance provides more protection by involving more physical qubits in the encoded system. The memory rate applies to stored information during correction cycles; it does not show a million general-purpose quantum gates or a complete fault-tolerant computation.

A separate two-qubit Clifford-gate measurement found an error-suppression factor of nine as the code distance rose from 3 to 5. That estimate carried uncertainty of +8 and −4.

The memory rate is roughly one-tenth the September Helix-code result. That earlier benchmark measured 4.6 × 10⁻⁵ errors per logical qubit per correction cycle, with an uncertainty range of +6.2 × 10⁻⁵ and −2.6 × 10⁻⁵. The studies used different codes on the same Helios hardware.

Quantinuum CEO Dr. Rajeeb Hazra said, “With this result, we've proven that we have the error-correction architecture and trapped-ion technology required to achieve universal fault-tolerance.”

Robin Blume-Kohout, co-director of Sandia National Laboratories’ Quantum Performance Lab, was not involved in the study. He said it showed “real, serious performance gains from quantum error correction—a logical qubit nearly 100x better than its component physical qubits.”

Quantinuum plans to deliver Apollo as a universal fault-tolerant computer by 2029. The company has not yet demonstrated below-threshold scaling for magic-state preparation.