As quantum computing systems become increasingly practical, they are also becoming available through cloud computing platforms. The benefit of this is that, like any other cloud computing resource, you can access it through your CloudBank account. In this solution, we outline the details you should consider as you explore the use of these systems through public cloud platforms.
There are roughly four components to consider when approaching cloud-based quantum computing:
- The cloud infrastructure and pricing policies around accessing quantum services.
- The quantum computer itself. Quantum computers are very much not all made equal, and different models tend to be optimized towards solving different sorts of problems. As such, most cloud platforms offer a diversity of machines.
- Quantum simulators used for free or low-cost testing of your quantum algorithms.
- The programming interface and libraries used to access the above (often specific to the cloud platform being used)
Vendors: Quantum Computer Comparison
The information in this table is current as of June 2026.
| Vendor | Cloud | Hardware Model | Qubit Count | Type of Quantum Computer | Pricing |
|---|---|---|---|---|---|
| IonQ | AWS Braket | Aria | 25 | Trapped‑ion (gate‑model) | $0.03 per shot (industry‑reported) |
| AWS Braket | Forte | 36 | Trapped‑ion (gate‑model) | $0.080 per shot + $0.300 per task | |
| Azure Quantum | Aria‑1 | 25 | Trapped‑ion (gate‑model) | $97.50 minimum per program (error mitigation on); $12.4166 minimum (off); per‑gate‑shot: $0.000220 (1‑qubit), $0.000975 (2‑qubit) | |
| Azure Quantum | Forte‑1 | 36 | Trapped‑ion (gate‑model) | $168.195 minimum per program (error mitigation on); $25.7899 minimum (off); per‑gate‑shot: $0.0001645 (1‑qubit), $0.001121 (2‑qubit) | |
| Azure Quantum | Forte Enterprise‑1 | 36 | Trapped‑ion (gate‑model) | Same as Forte‑1 pricing above | |
| Rigetti | AWS Braket | Aspen‑M‑3 | 79 | Superconducting (gate‑model) | $0.000425 per shot + $0.300 per task (Rigetti Cepheus pricing; Aspen‑M‑3 not separately priced in results) |
| AWS Braket | Ankaa‑3 | varies | Superconducting (gate‑model) | Not priced separately in results (Rigetti family uses Cepheus pricing) | |
| Azure Quantum | Provider listed; no model surfaced | — | Superconducting (gate‑model) | No pricing surfaced in results | |
| IQM | AWS Braket | Emerald | 54 | Superconducting (gate‑model) | $0.00160 per shot + $0.300 per task |
| AWS Braket | Garnet | 20 | Superconducting (gate‑model) | $0.00145 per shot + $0.300 per task | |
| QuEra | AWS Braket | Aquila | 256 | Neutral‑atom (analog Hamiltonian simulator) | $0.010 per shot + $0.300 per task |
| AQT | AWS Braket | IBEX Q1 | 12 | Trapped‑ion (gate‑model) | $0.02350 per shot + $0.300 per task |
| Pasqal | Azure Quantum | FRESNEL | 100 | Neutral‑atom (gate‑model) | Pricing not surfaced in search results (Azure notes provider‑controlled pricing) |
| Azure Quantum | FRESNEL_CAN1 | 100 | Neutral‑atom (gate‑model) | Same as above (no pricing surfaced) | |
| Quantinuum | Azure Quantum | Provider listed; no model surfaced | — | Trapped‑ion (gate‑model) | No pricing surfaced in results |
| IBM Quantum | IBM Cloud | Heron R1–R10 | 133 each | Superconducting (gate‑model) | Pay‑As‑You‑Go: $96/min; Flex: $72/min; Premium: $48/min |
| D‑Wave | AWS Marketplace | Advantage (Leap) | >5000 qubits | Quantum annealer (not gate‑model) | Pricing not published; Marketplace listing requires private offer / contact sales |
Types: Quantum Computer Comparison
The information in this table is current as of June 2026.
| Type of Quantum Computer | How It Works | Strengths | Limitations | Best For |
|---|---|---|---|---|
| Superconducting (Gate‑Model) IBM, Rigetti, IQM | Superconducting circuits cooled near absolute zero; microwave pulses implement gates. | Fast gate speeds; mature fabrication; strong error‑correction research; widely used in industry. | Shorter coherence times; limited qubit connectivity; scaling requires complex wiring. | General‑purpose algorithms, error‑correction research, fast circuit execution. |
| Trapped‑Ion (Gate‑Model) IonQ, Quantinuum, AQT | Individual ions suspended in electromagnetic traps; lasers manipulate qubits and entanglement. | Highest fidelities; long coherence times; all‑to‑all connectivity simplifies circuits. | Slower gate speeds; scaling to very large systems is challenging. | Algorithms sensitive to connectivity (VQE, QAOA), high‑precision circuits, early fault‑tolerance. |
| Neutral‑Atom (Gate‑Model or Analog) QuEra, Pasqal | Neutral atoms arranged in optical tweezers; Rydberg interactions create entanglement; arrays can be reconfigured. | Highly scalable (hundreds–thousands of qubits); flexible geometries; supports both gate‑model and analog modes. | Gate fidelities still improving; analog mode is powerful but less general‑purpose. | Quantum simulation, many‑body physics, large‑scale optimization, emerging gate‑model workloads. |
| Quantum Annealing D‑Wave | System evolves toward the lowest‑energy state of an Ising model; specialized hardware for optimization. | Thousands of qubits; excellent for certain optimization and sampling problems; mature commercial ecosystem. | Not a universal quantum computer; cannot run gate‑model algorithms (Shor, Grover, VQE, etc.). | Combinatorial optimization, graph problems, routing, scheduling, sampling. |
Learning Resources
- Learn Quantum Computation using Qiskit is a free general-purpose textbook published by IBM’s quantum group that teaches the basics of quantum computation and algorithms using their qiskit domain-specific language
- Azure announces topological qbit progress
