Skip to content
EVIDENCE RECORD

Google Quantum AI: what changed.

The recorded research and its supporting evidence.

This is a historical record. Read the current company research →
  1. Evidence or assessment clarified · axis ratings unchanged

    Assessment reviewed 6 Sept 2026 · Research reviewed 13 Oct 2026
    What changed
    Research evidence: Quantum error correction below the surface code threshold
    Previously

    Peer-reviewed Google-authored experiment demonstrates below-threshold memories and a logical lifetime exceeding its best physical qubit. The current article incorporates an April 2026 figure-label correction.

    Source title
    Quantum error correction below the surface code threshold
    Source publication date
    9 Dec 2024
    Source type
    Peer-reviewed vendor research
    Marked independent
    No
    Now

    Peer-reviewed Google-authored experiment demonstrates below-threshold memories and a logical lifetime exceeding its best physical qubit. The current article incorporates an April 2026 figure-label correction.

    Source title
    Quantum error correction below the surface code threshold
    Source publication date
    9 Dec 2024
    Source type
    Peer-reviewed vendor research
    Marked independent
    No
    Where to look
    Abstract; main memory/decoder distinction; Change history
    Source checked
    12 Nov 2026
    Source or product version
    Published 9 December 2024; correction 28 April 2026
    Limitations
    Memory and real-time decoder experiments are distinct; not a complete useful fault-tolerant algorithm.
    Retained for this claim (1)
    Research evidence: Programmable digital quantum simulation of 2D Fermi Hubbard dynamics using 72 superconducting qubits
    Previously

    Phasecraft-led preprint reports 72-qubit Willow experiments, with exact classical verification restricted to tractable cases; it is research-partner evidence rather than independent replication of Google's memory result.

    Source title
    Programmable digital quantum simulation of 2D Fermi-Hubbard dynamics using 72 superconducting qubits
    Source publication date
    30 Oct 2025
    Source type
    Research-partner preprint
    Marked independent
    No
    Now

    Phasecraft-led preprint reports 72-qubit Willow experiments, with exact classical verification restricted to tractable cases; it is research-partner evidence rather than independent replication of Google's memory result.

    Source title
    Programmable digital quantum simulation of 2D Fermi-Hubbard dynamics using 72 superconducting qubits
    Source publication date
    30 Oct 2025
    Source type
    Research-partner preprint
    Marked independent
    No
    Where to look
    Abstract; submission history
    Source checked
    12 Nov 2026
    Source or product version
    arXiv:2510.26845v3, 23 September 2026
    Limitations
    Exact checks are restricted to tractable regimes; research-partner experiment.
    Retained for this claim (1)
    Research evidence: Willow Early Access Program
    Previously

    Explicitly states hardware is not public and that selected applicants have been notified after the May 2026 deadline.

    Source title
    Willow Early Access Program
    Source publication date
    Not recorded
    Source type
    Official access program
    Marked independent
    No
    Now

    Explicitly states hardware is not public and that selected applicants have been notified after the May 2026 deadline.

    Source title
    Willow Early Access Program
    Source publication date
    Not recorded
    Source type
    Official access program
    Marked independent
    No
    Where to look
    Program introduction; deadline and selection status
    Source checked
    12 Nov 2026
    Source or product version
    Living page checked 12 November 2026
    Limitations
    Selected cohort; no public Willow entitlement.
    Retained for this claim (1)
    Research evidence: Access and authentication — Cirq
    Previously

    Hardware access requires an approved research group, approved project and authorized users; installing Cirq alone does not grant hardware access.

    Source title
    Access and authentication — Cirq
    Source publication date
    Not recorded
    Source type
    Vendor technical documentation
    Marked independent
    No
    Now

    Hardware access requires an approved research group, approved project and authorized users; installing Cirq alone does not grant hardware access.

    Source title
    Access and authentication — Cirq
    Source publication date
    Not recorded
    Source type
    Vendor technical documentation
    Marked independent
    No
    Where to look
    Access and authentication; API Access; User access
    Source checked
    12 Nov 2026
    Source or product version
    Page last updated 22 July 2026; checked 12 November
    Limitations
    Approved groups/projects/users are required; installing Cirq is insufficient.
    Retained for this claim (1)
    Sources used in this record (5)

Recorded dates use UTC and show when QubitWire saved the comparison. Review dates show the evidence check; neither is a new claim about when the underlying event happened. This view preserves the original research and assessment methodology.