Skip to content
Home/Quantum Computing News/Quantum Processors by Qubit Count: The Machines Running in 2026
HardwareIndustry

Quantum Processors by Qubit Count: The Machines Running in 2026

A current list of quantum processors from IBM, Google, Atom Computing, QuEra, D-Wave, Rigetti, IonQ, and Quantinuum, with qubit counts. The list explains why raw qubit count misleads and what to compare instead.

FreeQuantumComputing
·· 7 min read

Qubit counts make headlines, and they mislead more than any other number in quantum computing. This list gives the current machines and their counts, then explains why the count is the least useful column in the table.

Read the physical vs logical split first

A physical qubit is one real qubit in the machine. A logical qubit is an error-corrected unit built from many physical qubits, and it is the one that holds up through a computation. Comparing a 5,000-qubit annealer to a 36-qubit trapped ion machine compares two different things. See Logical Qubits and Fault Tolerance and What a 100-Qubit Machine Means for the full distinction.

The current machines

ProcessorQubitsModalityCompanyYear
Advantage25,000+AnnealingD-Wave2025
(unnamed)1,180Neutral atomAtom Computing2024
Condor1,121SuperconductingIBM2023
Aquila256Neutral atomQuEra2022
Radiance150SuperconductingIQM2025
Heron133SuperconductingIBM2023
Willow105SuperconductingGoogle2024
Ankaa-384SuperconductingRigetti2024
Helios48 logicalTrapped ionQuantinuum2025
Forte Enterprise36 algorithmicTrapped ionIonQ2024

The list changes quickly. Vendors ship new processors roughly yearly, and qubit counts rise on every cycle. Check the top companies list for current rankings.

Why the count misleads

Three machines with the same physical qubit count deliver different amounts of work. A trapped ion machine holds state for seconds and connects every qubit to every other. A superconducting machine runs fast gates but forgets in a hundred microseconds. An annealer holds the most qubits of all and cannot run a general circuit. The count says none of this.

Two better numbers matter. Fidelity sets how many gates you run before the answer is noise. Quantum volume bundles size, fidelity, and connectivity into one benchmark. Neither makes a headline, both predict performance. See Quantum Volume: How to Measure It Yourself.

What to compare instead

Ask three questions of any processor. How many logical qubits does it produce, and when? What is the two-qubit gate fidelity? How much of the hardware is error-corrected today? A machine with a small count and real logical qubits, like Helios at 48, sits closer to useful work than a machine with a big count and no correction at all.

Next steps