Quantum Benchmarks Explained: Why Qubit Count Misleads
Why raw qubit count misleads
Imagine two machines. One has many qubits that make errors in nearly every operation. The other has far fewer qubits that are much more accurate. For many programs the smaller, better machine gives a usable answer, and the larger one returns mostly noise. A big number on a press release says little about whether qubits are good enough to chain together into a long circuit.
The basic quality measures
- Gate error rate and fidelity: fidelity is how closely an operation matches the ideal one, and error rate is roughly what is left over. Two qubit gates are usually worse than single qubit gates, and they are often the number to watch. Small differences, like 99.5 percent versus 99.9 percent, matter a lot because errors multiply over thousands of gates.
- Readout error: how often measurement reports the wrong value.
- Coherence time: how long a qubit keeps its quantum state before noise destroys it, compared with how long gates take.
- Connectivity: which qubits can interact directly. Poor connectivity forces extra swap operations, which add errors.
- Speed: how many circuit layers per second can be run.
Composite benchmarks
Because single numbers hide details, researchers use tests that stress the whole system. Randomized benchmarking applies long random gate sequences to estimate average gate error. Quantum volume, introduced by IBM, tests the largest random square circuit a machine can run with acceptable success, and it blends qubit count, error rates and connectivity into one figure. Cross entropy benchmarking compares output patterns with a classical simulation, and was used in supremacy experiments. Newer application based suites run small versions of real algorithms to see how a machine performs on practical tasks.
Limits of benchmarks
No single score is complete. Quantum volume can be gamed by tuning for that specific test and it says little about structured algorithms. Random circuit tests may not reflect real workloads. Results also depend on calibration at the moment of the test, so numbers can vary day to day and between a vendor's best chip and its average one. Companies choose which results to publish, so independent comparisons are valuable.
It also helps to know that error rates are often reported as averages over many operations, while the worst qubit pair on a chip can be noticeably worse. For a long circuit, the weakest link matters, so a good report shows distributions and not only the best number.
What to look for in a claim
- Is the metric defined, and how was it measured?
- Is it a best case or an average across the chip?
- Does it describe physical qubits or error corrected logical qubits? See error correction.
- Has anyone independent reproduced it?
What is still unknown
The field has not agreed on one standard measure of usefulness, and the right yardstick may change once machines are error corrected. Until then, treat headlines with the same caution described in quantum computing myths.
Frequently asked questions
Why does qubit count not tell you how good a quantum computer is?
Because noisy qubits ruin long calculations. Error rates, connectivity and coherence determine how much a machine can actually do.
What is fidelity?
It measures how closely a real operation or state matches the ideal one. Higher is better, and small gaps compound over many gates.
What is quantum volume?
It is a benchmark introduced by IBM that scores the largest random circuit a machine can run well, combining qubit count, error rates and connectivity.
What is coherence time?
It is how long a qubit holds its quantum state. What matters is how many gate operations fit inside that time.
Can benchmarks be misleading?
Yes. Scores can reflect best case conditions or tuning for the test, so it helps to look for independent reproduction.
Are logical qubits counted the same as physical qubits?
No. A logical qubit is built from many physical qubits using error correction, so the two numbers should never be compared directly.
Keep reading
- NISQ Explained: Noisy Intermediate-Scale Quantum Computers
What does NISQ mean? Learn why today's noisy, mid-size quantum computers are limited, what they can do, and how the field plans to move past them. - Quantum Error Correction Explained
Qubits are fragile, so quantum computers need error correction. Learn how logical qubits are built and why this is the key challenge. - Quantum Supremacy vs Quantum Advantage Explained
What the terms quantum supremacy and quantum advantage mean, and why headline claims are often debated. - Types of Quantum Computers: Superconducting, Ion, Photonic and More
A guide to the main ways quantum computers are built, with the strengths and trade-offs of each approach.
All Quantum computing guides | Back to top | Search the site
Main pages: Quantum computing explained | Quantum and crypto | Companies | Quantum news | Glossary