Hardware

Real quantum hardware

Everything else on this site is exact simulation. On this page a circuit leaves the browser, runs on a physical processor and comes back with the noise that simulation does not have. The lane never shows a device result that did not come from a device.

Two things work today: the exact prediction, and the comparison below against a real device's published calibration noise model. The live run switches on with a token, and the status panel says where it stands. Read the hardware scoreboard for what these devices have and haven't achieved.

Lane status

Checking…

checking

What activates it: deploy the FastAPI service (one-click Render blueprint in the backend repo) and set QISKIT_IBM_TOKEN there, then NEXT_PUBLIC_TRANSPILER_API_URL here. The guardrails (size caps, a monthly shot ledger, one job at a time) are enforced on the server and covered by the backend's tests.

Submit a circuit

OpenQASM 2.0, up to 5 qubits. Build one in the Circuit Lab and paste its export here.

Simulation vs. device: CHSH

CHSH Bell test on a 2-qubit Bell state, 4 measurement settings. Classical bound 2, Tsirelson bound 2.83.

ideal statevector (noiseless)S = 2.828
ibm_brisbane noise model (Aer, real calibration snapshot)S = 2.444

Simulation using the published error rates of a real 127-qubit device. Not a hardware run.

real hardwarenot run

Not run yet. Set QISKIT_IBM_TOKEN and re-run this script to execute on an IBM Quantum device.

Amber tick = classical bound. Generated 2026-08-22 by scripts/hardware/ibm_noise_model.py (Qiskit Aer + IBM calibration snapshot).