Superconducting

IBM Quantum

The world's largest superconducting quantum computing platform. Free cloud access, open-source Qiskit SDK, and 100+ organizational partners in the IBM Quantum Network.

  • 156 qubits (Heron r2)
  • ~99.9% gate fidelity
  • Free 10 min/month
  • Qiskit SDK

Public quantum computing, pioneered since 2016

IBM launched the IBM Quantum Experience in 2016, becoming the first company to offer public cloud access to real quantum hardware. Since then, IBM has released a steady cadence of increasingly powerful systems: the 27-qubit Falcon in 2019, the 127-qubit Eagle in 2021, the 433-qubit Osprey in 2022, and the milestone 1,121-qubit Condor in 2023. The current flagship system for production workloads is IBM Heron r2, a 156-qubit processor using direct CNOT gates that achieve median two-qubit fidelity above 99.9%.

IBM's superconducting qubits are transmon-type: small Josephson junction circuits cooled to millikelvin temperatures in dilution refrigerators. IBM arranges these on a heavy-hex lattice, a 2D connectivity graph where each qubit connects to at most three neighbors. This sparse connectivity limits native SWAP-free routing but allows longer coherence times and lower error rates than denser topologies.

The primary SDK for IBM hardware is Qiskit, an open-source Python framework that spans circuit construction, transpilation, execution via the Sampler and Estimator primitives, and post-processing. Qiskit Patterns provides a structured programming model for near-term algorithms: map, optimize, execute, and post-process. For hybrid workloads, Quantum Serverless extends Qiskit jobs into a classical-quantum compute cluster model.

System specs at a glance

Specification Value
Current systems IBM Heron r2 (156q), IBM Heron r1 (133q), IBM Eagle (127q), IBM Condor (1121q, research only)
Qubit technology Superconducting transmon qubits
Two-qubit gate fidelity Up to ~99.9% on best qubit pairs (Heron r2)
Coherence time (T1/T2) 100–500 microseconds
Connectivity Heavy-hex lattice (not all-to-all)
Two-qubit gate time ~100–400 nanoseconds
Throughput metric CLOPS (up to ~340K circuit layer operations/sec on Heron r2)
Cloud access IBM Quantum Platform, IBM Cloud (pay-per-use)
Primary SDK Qiskit (open source, Python)
Supported SDKs Qiskit, PennyLane-qiskit plugin, Cirq (via transpilation), OpenQASM 3
  • IBM Heron r2

    156 qubits

    IBM's current flagship processor. The Heron r2 improves on the original 133-qubit Heron with direct CNOT gates and median two-qubit fidelity above 99.9%. IBM's recommended system for production algorithm workloads requiring the best gate quality.

  • IBM Eagle

    127 qubits

    The 127-qubit Eagle systems are available on the free Open Plan (10 min/month). Introduced in 2021, Eagle was the first quantum processor to surpass 100 qubits. Multiple Eagle systems remain the primary free-tier hardware.

  • IBM Condor

    1,121 qubits

    A 2023 research milestone demonstrating IBM's fabrication scale. Condor is a proof-of-concept for large-scale integration and is not a primary production system. Its achievement informed subsequent architectural decisions for Heron.

  • IBM Falcon

    27 qubits

    The workhorse of earlier IBM Quantum generations. 27-qubit Falcon systems have served millions of user jobs and remain useful for small-scale experiments and educational circuits. Some are still accessible via IBM Quantum.

Where IBM Quantum hardware excels

  • Drug discovery and molecular simulation

    Qiskit's chemistry modules and the Estimator primitive enable variational quantum eigensolver (VQE) calculations for molecular ground-state energies at scales approaching chemical accuracy.

  • Materials simulation

    Heavy-hex lattice connectivity maps well to 2D lattice problems in condensed matter physics, making IBM hardware a strong fit for Hubbard model and spin-chain simulations.

  • Finance and Monte Carlo

    Quantum amplitude estimation offers a quadratic speedup over classical Monte Carlo for pricing derivatives and estimating risk, a use case IBM has actively demonstrated with financial partners.

  • Combinatorial optimization

    QAOA on IBM hardware with Qiskit Patterns delivers structured approaches to MaxCut, portfolio optimization, and logistics problems using the Sampler primitive.

  • Quantum machine learning

    Parameterized quantum circuits implemented via Qiskit Machine Learning integrate with classical ML pipelines for classification, generative modeling, and kernel methods.

  • Error mitigation research

    IBM's Estimator primitive has built-in zero-noise extrapolation (ZNE) and probabilistic error amplification (PEA), making IBM hardware ideal for near-term error mitigation experiments.

Run your first circuit on IBM Quantum

  1. Create a free IBM Quantum account

    Go to quantum.ibm.com and sign up with a free IBM account. The Open Plan gives you 10 minutes of quantum compute time per month on 127-qubit Eagle systems at no cost. No credit card required.

  2. Install Qiskit and the IBM Runtime client

    pip install qiskit qiskit-ibm-runtime

    qiskit installs the core SDK for circuit construction and transpilation. qiskit-ibm-runtime provides the client for authenticating with IBM Quantum and submitting jobs via the Sampler and Estimator primitives.

  3. Save your API token

    from qiskit_ibm_runtime import QiskitRuntimeService
    
    QiskitRuntimeService.save_account(
        channel="ibm_quantum",
        token="YOUR_API_TOKEN",  # from quantum.ibm.com account page
        set_as_default=True,
    )

    Your API token is found on your IBM Quantum account page. Running save_account stores it locally so you do not need to pass it on every run.

  4. Build and submit a Bell state circuit

    from qiskit import QuantumCircuit
    from qiskit_ibm_runtime import QiskitRuntimeService, SamplerV2 as Sampler
    
    service = QiskitRuntimeService()
    backend = service.least_busy(operational=True, simulator=False, min_num_qubits=2)
    
    qc = QuantumCircuit(2)
    qc.h(0)
    qc.cx(0, 1)
    qc.measure_all()
    
    sampler = Sampler(backend)
    job = sampler.run([qc], shots=1024)
    result = job.result()
    print(result[0].data.meas.get_counts())

    This selects the least busy available IBM Quantum backend and submits a Bell state measurement. Results come back as bitstring counts.

  5. Explore IBM Quantum Learning

    IBM Quantum Learning at quantum.cloud.ibm.com offers free structured courses covering Qiskit basics, quantum algorithms, and error mitigation. Labs within the courses run directly against IBM hardware using your account's compute time.

Access plans and costs

IBM Quantum offers multiple access tiers from a generous free plan to premium institutional membership. Most learners start with the Open Plan at no cost.

  • IBM Quantum Open Plan

    Free (10 min/month)

    Free access to 127-qubit Eagle systems. Each account receives 10 minutes of quantum compute time per month. No credit card required. Best for learning and small experiments.

    View details →
  • IBM Quantum Learning

    Free educational access

    IBM Quantum Learning provides free courses, guided labs, and limited hardware access tied to course exercises. No account needed to read materials; IBM account needed to run circuits.

    View details →
  • IBM Quantum Network

    Academic / commercial membership

    100+ universities and companies have priority access to IBM's full system fleet, including Heron and premium systems, via the IBM Quantum Network. Apply through IBM.

    View details →
  • IBM Cloud Pay-per-Use

    Per second of runtime

    Access IBM Quantum hardware via IBM Cloud with a pay-per-use model billed per second of quantum runtime. Suited for production workloads and teams needing repeatable, scalable access.

    View details →

Tutorials and reference docs