Applications / AI / Quantum for AI
QML and classification
Labelling data through a quantum feature map
Two methods, one dataset. A variational quantum classifier trains rotation angles the way a network trains weights and reads a label off the measurement. A quantum kernel trains nothing in the circuit at all: it embeds each point with a fixed feature map, measures the overlap of every pair, and hands the resulting matrix to a classical support vector machine. The kernel is the method most of the early advantage claims were made with.
Both were run on the same two-moons split with the same three seeds as four classical baselines, so the comparison is like for like. Two moons is low-dimensional classical geometry, which is where a quantum feature map is least expected to help: the separations proved in the literature are on datasets constructed for the purpose.
Run on our engines
Two moons, 22 training and 40 test points, seed 1 to 3, against logistic regression, k-NN, an MLP and an RBF kernel on the identical split. Submitted to each kind of compute we offer, on 17 September 2026. Every figure below is a real job on the service, priced as any customer would be priced.
| Device | Engine | Kind | Qubits | Result | Cost |
|---|---|---|---|---|---|
 | exact.cpu | CPU | 2 | variational classifier 0.925 / 0.825 / 0.700 across three seeds | $0.4000 |
| exact.cpu | CPU | 2 | quantum kernel 0.475 / 0.700 / 0.700, at 1,111 circuits a seed | $0.1111 |
The same problem is yours to run: every instance here is seeded, so it rebuilds exactly. Open the console and a cost estimate is free before anything executes.
The classical baselines on the same split return 0.900 / 0.775 / 0.900 for logistic regression, 0.950 / 0.875 / 0.950 for k-NN and 0.875 / 0.900 / 1.000 for the MLP. The RBF kernel reaches 0.850 untuned.
The kernel's first seed returned 0.475, below chance on a balanced problem, so it was checked rather than reported. Computing the same kernel exactly with no shots returns 0.475 as well, and the overlap of a point with itself is 1 to fifteen decimal places, so 1,111 measured circuits reproduced the noiseless calculation.
Every quote matched its charge exactly at $0.1111, three times, on a real 1,111-circuit workload.
Where this stops
- The kernel result turns on how far the input is scaled before it is written into rotation angles, a constant carrying no information about the problem: the same map and data swing from 0.400 to 0.825 across four settings
- A kernel needs one circuit per pair of points, so circuit count grows with the square of the dataset
- Tuning one side of a comparison and not the other overstates the result in either direction, so the classical baselines here are the untuned defaults
The classical baseline for this case is logistic regression, k-NN, an MLP and an RBF kernel. The structural limit, which does not move when a benchmark is re-run, is A kernel needs one circuit per pair: 1,111 circuits for 62 points.
Run it yourself
Every figure above is from a job billed on the production service. Load the same circuits into the console, change the instance to your own data, and export a certificate for your own run rather than citing ours.
The argument behind these numbers, at length: Quantum machine learning on simulators and real quantum hardware.
Other AI use cases
QCBM and quantum generative models
Sampling from a distribution learned from scratch
QCBM against its own untrained start
QGAN and synthetic data
Adversarial generation against a learning discriminator
QGAN with a classical discriminator
Quantum AI image generation
Writing an image into a circuit and reading it back
Patch GAN and NTQIP, both on Rigetti