Benchmark

Surface Code Quantum Error Correction Below Threshold

Distance-5/7 surface code memory · Other · 105 qubits · Google Quantum AI stack (Cirq-compatible)

Memory benchmark demonstrating a quantum error-correcting code operating below the fault-tolerant threshold on a superconducting processor. Google's team ran distance-3, -5 and -7 surface codes and showed that logical error rate per cycle drops by a factor Lambda each time code distance is increased by two, proving exponential error suppression with scale. Also ran a distance-29 repetition code. The distance-7 logical memory outlived its best physical qubit by a factor of 2.4, a first-of-its-kind result for superconducting hardware.

error-correctionsurface-codebelow-thresholdlogical-qubitsuperconducting

Memory benchmark demonstrating a quantum error-correcting code operating below the fault-tolerant threshold on a superconducting processor. Google's team ran distance-3, -5 and -7 surface codes and showed that logical error rate per cycle drops by a factor Lambda each time code distance is increased by two, proving exponential error suppression with scale. Also ran a distance-29 repetition code. The distance-7 logical memory outlived its best physical qubit by a factor of 2.4, a first-of-its-kind result for superconducting hardware.

Key Metrics
Error per cycle (d=7)
0.143 ± 0.003%
Lambda (suppression factor)
2.14 ± 0.02
Logical vs physical lifetime
2.4 ± 0.3×
Decoder latency (d=5)
63microseconds
Cycle time
1.1microseconds
Why It Matters

First demonstration on superconducting hardware that increasing code distance reduces logical error exponentially — crossing the practical break-even point required for scalable fault-tolerant quantum computation.

Hardware

Google superconducting processor (Willow-generation, 105 qubits)

Framework

Google Quantum AI stack (Cirq-compatible)