Google tests quantum contextuality with as many as 105 qubits
The experiments outperformed classical strategies under specified conditions, but researchers did not claim a general advantage over classical computing.

Google Quantum AI says experiments with its superconducting processors demonstrated contextuality across systems containing as many as 105 qubits. The research, featured on the cover of Science, also identified hardware errors while stopping short of establishing an unconditional advantage over classical computers.
Contextuality describes how the outcome of a measurement depends on the other measurements performed with it, rather than being predetermined. Previously, demonstrations had largely involved only a few particles. The property is thought to underpin some quantum algorithms.
For the two-player Mermin-Peres magic square game, Google's qubits achieved a 98.3% win rate, compared with a maximum of 88.9% for a classical strategy. A separate task involving 105 qubits outperformed classical approaches operating with limited resources. However, the quantum advantage in one game diminished as the number of participating qubits increased.
Google says the tests can detect hardware faults missed by other benchmarks and could support future chemistry and materials science applications. Neither field was simulated in these experiments. The work initially appeared in preprint form in December 2025.




