IBM Brisbane
Superconducting QPU (Eagle r3, 127 qubits) accessed via Qiskit Runtime. Historical data from February–June 2025, collected by Sami. Future runs are automated monthly.
Consistency over time
Within-run standard deviation per run — the primary stability metric for this benchmark. Lower is more consistent.
Success probability over time
Success probability for a given circuit is the fraction of shots that produced the correct output — where "correct" is the deterministic, noise-free answer computed by classical simulation. Each point is the mean across the circuits sampled that run. The shaded band shows ±1 standard deviation within the run.
Performance breakdown
How success probability varies across circuit depth and input state, aggregated across all runs.
Success probability by circuit depth and input state
Each point is one (depth, input state) combination. Point size reflects how many circuits were run with that combination. Drag to rotate.
Distribution by circuit depth
Mean success by circuit depth
Mean success probability for each depth, averaged across all runs. A declining trend confirms that noise accumulates as circuit depth increases.
Mean success by input state
Does the initial qubit state affect results? Ideally it shouldn't — deviations suggest state-preparation or readout asymmetry.