Quantinuum Claims 99.9995% Logical Fidelity

In brief

Quantinuum (19 October 2026) says its Helios commercial computer achieved a logical fidelity of 99.9995% — “just five errors per million operations” — using a colour code.
For the fault-tolerance race, the company calls this a world record: 10× lower error than its own September result and 100× lower than the best published result on other modalities (attributed to Google Quantum AI on superconducting hardware). Those rankings are Quantinuum’s claims.
What’s still unknown: whether the result scales to larger logical computation and commercial use — the release itself flags replication risk — and whether Apollo by 2029 arrives on that company forecast.

Trapped-ion Helios is the same machine that hosted Quantinuum’s September Helix-code demo; the company says switching codes inside one architecture is the point.

What did Quantinuum measure?

The release says Helios achieved 99.9995% logical fidelity with the company’s error-correction architecture. It also says each tested increase in code distance cut remaining logical errors by a factor of 12.

Claim (company)	Figure
Logical fidelity	99.9995%
Errors	5 per million operations
vs own Sept record	10× lower error
vs best other modality	100× lower (Google Quantum AI, superconducting)
Code	Colour code
Physical qubits / logical	Up to 37 (colour); Helix used 10

Source: Quantinuum press release, 19 October 2026. Record and comparisons are company claims.

What did an independent researcher say?

Robin Blume-Kohout of Sandia National Laboratories’ Quantum Performance Lab, who was not involved in the study, said the paper shows “a logical qubit nearly 100x better than its component physical qubits,” and that a scaled version “could reach the reliability required for useful fault-tolerant computation at a feasible size—less than a million physical qubits.”

What about Apollo?

Quantinuum says the result puts accuracy “within the target range for Apollo,” its planned universal fault-tolerant machine “expected to be available by 2029.” That timeline is a company forecast, not a shipping date.

CEO Rajeeb Hazra said the team had “reduced error rates tenfold in rapid succession using a different code on the same commercial hardware.”

Who is affected?

Quantum algorithm and materials teams watching logical error floors
Competitors on superconducting and other modalities named in the comparison
Investors tracking Quantinuum’s path to Apollo

Our read

A second logical-memory leap in two months on one commercial Helios box is a strong trapped-ion signal — especially if the Sandia reading of “nearly 100× better than physical” holds under scrutiny. The catch is the company’s own forward-looking caveat: experimental settings may not replicate at larger scale or in logical computation. Treat “world record” and the Google 100× line as Quantinuum-framed, and wait for peer review and independent replication before calling the FTQC race settled.

What’s still unknown?

Peer-reviewed publication detail and independent reproduction
Logical computation (not only memory) at this fidelity
Whether Apollo’s 2029 window holds

How we sourced this

Sole primary: Quantinuum press release, 19 October 2026.
“World record” and 10× / 100× comparisons attributed as company claims. Blume-Kohout quote taken from the same release.