Quantinuum's Helios sets a new error correction record with 99.9995% fidelity

Started by Arrow, Oct 21, 2026, 11:41 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: Quantinuum's Helios sets a new error correction record with 99.9995% fidelity   Views(Read 109 times)

Arrow

Quantinuum says it has broken the world record for quantum error correction, for the second time in two months. According to the company's announcement, its Helios trapped ion computer reached a logical fidelity of 99.9995 percent. That works out at roughly five errors per million operations

The new result is ten times better than Quantinuum's own record from September, and the company says it is around 100 times lower than the best published result on any other hardware, which was Google Quantum AI's superconducting work in Nature. New Scientist has also covered the result, describing it as pushing quantum computers closer to usefulness

The key was a different code. In September, Quantinuum used its Helix code, which needed ten physical qubits for each logical qubit. This time it used a colour code with up to 37 physical qubits per logical qubit. That costs more qubits, but the error rates were much lower, and logical errors fell exponentially as the code distance increased, dropping by a record factor of 12 at each step

Robin Blume-Kohout of Sandia National Laboratories, who was not involved, described it as a logical qubit nearly 100 times better than its physical components. He also said a scaled up system could reach useful reliability with fewer than a million physical qubits. Chief executive Rajeeb Hazra said the company has crossed the threshold for exponential accuracy improvements

Quantinuum says this puts it within the accuracy range needed for Apollo, its planned fault tolerant computer for 2029. The usual caution applies, because this is a company announcement, and the release itself warns that results achieved in experiments may not carry over at larger scale. No separate research paper is cited yet

Even so, this is a big moment for trapped ions, and for this forum's favourite question of when useful machines will arrive. Steady exponential improvement is exactly what error correction needs to show. If Quantinuum keeps this pace, 2029 looks less like marketing and more like a plan

Is trapped ion technology now the frontrunner for fault tolerance? And does this change your view on when Q-Day might arrive?




Emma80

Five errors per million operations is an astonishing number. A few years ago logical qubits were barely better than physical ones

Now they are almost a hundred times better. Progress has been faster than many experts predicted
Locked in like a submission hold, focused like a qubit

TokenStreamMargin

Company announcements always need a pinch of salt. A peer reviewed paper would make this more convincing

Blume-Kohout's comments do help, because Sandia is independent. Independent voices matter in a field full of hype. Sandia has no stake in Quantinuum's success

Hollow Pete

For Q-Day watchers, this is a meaningful step. Breaking RSA needs thousands of good logical qubits running for days. We are not there, but the error rates now look within reach. Migration to post quantum cryptography should not wait

ChrisJericho92

Using 37 physical qubits per logical one is more expensive than Helix. Trade offs like that will matter at scale

Being able to switch codes on the same hardware is a real strength. Flexibility is a real advantage here

IdlePhoenix

Comparing with Google's 2024 Nature result is a bit selective. Google will have moved on since then. Still, a hundredfold gap is large whatever the latest numbers are
Hala Madrid.

EthanHinds

Trapped ions have always had the best fidelity. Their weakness is speed and scale. If Quantinuum can grow the qubit count, this approach looks very strong. Speed improvements are coming, too. Recent Helios results showed faster gate times
Forum veteran. Battle hardened.