Google Quantum AI Demonstrates Contextuality Across 105 Qubits

What happened
On November 19, 2026, Google Quantum AI announced the publication of research in Science demonstrating measurement contextuality using superconducting quantum processors with up to 105 qubits. The experiments included a coordination game where the quantum processor achieved a 98.3% win rate, exceeding the classical maximum of 88.9%. The work extended previous contextuality tests, which had been limited to small systems, to much larger quantum devices. The research also introduced new hardware diagnostics based on contextuality, revealing subtle errors not detected by other tests.
Key details
- Measurement contextuality was demonstrated on processors with up to 105 qubits.
- A Kochen–Specker–Bell test produced a result of χKSB = 5.618 ± 0.005, surpassing the classical bound of 4.
- Tests included a GHZ parity game across up to 71 qubits, with quantum advantage observed at all tested sizes.
Why it matters
Researchers and companies working on quantum computing now have a way to test quantum hardware at large scales using contextuality-based tasks, which can reveal errors that other benchmarks may miss. This provides a more precise method for evaluating and diagnosing quantum processors as they grow in size. The findings mean that as quantum devices scale up, maintaining their computational advantage becomes more challenging, and new diagnostic tools are needed to ensure reliable operation.
