Quantum Chip Beats Classical Limit in New Test
Google's 105-qubit Willow result is a research-stage benchmark, and rival labs will be measured against its scale.
Updated on Nov. 20, 2026 in Quantum Computing

Google Quantum AI's 105-qubit Willow processor won the four-qubit Mermin-Peres magic square game at a 98.3% success rate, against a classical strategy ceiling of 88.9%. The result, published in Science on November 19, 2026, is a research-stage demonstration on a superconducting chip, not a commercially available product.
Why it matters
The experiment tests whether a quantum machine can beat any classical strategy on a task built around quantum contextuality, a property with no classical equivalent. Running the test across all 105 qubits signals that the field is moving from small proofs of principle toward larger, more reliable quantum hardware.
The evidence comes from a peer-reviewed experiment in which Willow ran nine game assignments, each executed 100,000 times, and scored 98.3% against a classical ceiling of 88.9%. The team has not reported how much of the gap would close with lower hardware noise, which the authors attribute as the main cause of the remaining errors.
The players
Google Quantum AI
Google's quantum research division, which builds superconducting qubit processors and announced the 105-qubit Willow chip in December 2024.
The details
Quantum contextuality is a property in which the outcome of a measurement depends on which other measurements are performed alongside it, something classical physics cannot reproduce. The Mermin-Peres magic square game uses this property: two players cooperate on a grid of measurements to satisfy rules that no fixed classical strategy can meet every time. Willow, a superconducting processor whose qubits are tiny circuits cooled to near absolute zero, executed the quantum version of the game and exceeded the classical limit. The study extended the same contextuality tests to multi-player GHZ-type games and to hidden linear function problems, the latter using a shallow-circuit algorithm.
Timeline
December 2024: Google Quantum AI announced the Willow chip.
December 2025: A preprint version of the study appeared on arXiv.
November 19, 2026: Science published the peer-reviewed study, featuring the work on its cover.
The Tech Race
The result positions Willow against rival superconducting and other quantum efforts, which the source notes will likely be measured against the scale of this demonstration. It moves the competition from raw qubit counts toward demonstrated tasks that classical machines cannot match.
This is a research demonstration on a chip that is not sold to the public, so there is no product or pricing to act on. Practical effects will reach users only once quantum processors become reliable enough for commercial workloads, a timeline the source does not give.
The takeaway
The trajectory is that quantum hardware is being tested on tasks that classical computers provably cannot match, and the next signal is whether rivals publish comparable results at comparable scale. Watch for other labs reporting contextuality or GHZ-type game results on their own hardware.
Further reading
For background on the hardware behind these results, see the Quantum Computing coverage.


