Quantinuum says its 99.9995% logical fidelity on Helios is not comparable to Google's magic-state result
Quantinuum says Google's 99.99% result measures magic-state preparation, while its own result measures a logical computation. It claims higher fidelity and 85% retention on a comparable test.
What happened
Quantum computers make many errors. Error correction groups physical qubits into one more reliable logical qubit. Quantinuum says its logical operations on its Helios machine are right 99.9995% of the time, or five errors per million operations.
Why it matters to you
Companies are publishing record error numbers that look alike but measure different things. Anyone judging quantum progress needs to know what each number actually covers.
The takeaway
Two record error figures from Quantinuum and Google measure different steps, so they cannot be compared head to head.
What it means for you
Compare logical computation results against distance-scaling work, and compare magic-state results against other magic-state work, including retention rate and qubit overhead.
Quantinuum argues customers buy answers, not gate fidelities, so resource efficiency matters for the hardware needed to run useful workloads.
This is a company blog post, and Quantinuum's claims of leadership and its 2029 Apollo target are forward-looking statements it says carry risks.