Retrospectively entered on 24 September 2026. Timeline date reflects the publication date of the underlying research.
What happened
On 13 May 2026, Nature published the Jiuzhang 4.0 team’s paper on Gaussian boson sampling with 1,024 squeezed states in an 8,176-mode programmable photonic circuit. The researchers report detection events involving up to 3,050 photons, 92% source efficiency and 51% overall system efficiency. They compare the results against then-current classical simulation methods.
Evidence and limitations
The specifications and benchmarking claims are taken from the peer-reviewed study and cross-checked against the Chinese Academy of Sciences’ account and subsequent Nature Physics commentary. Gaussian boson sampling is a specialised benchmark; the reported results do not demonstrate a general-purpose fault-tolerant quantum computer, useful cryptanalysis or superiority on arbitrary practical problems. Classical simulation methods can improve after publication.
Assessment
The result documents progress in large-scale photonic state generation, detection and programmable optical circuits. Assessments of computational advantage must retain the exact task, hardware configuration and classical benchmark used.
Why it matters
Photonic and superconducting systems face different scaling constraints. Recording them separately avoids treating photon counts, optical modes and physical qubits as interchangeable measures.
Sources and cross-checks
- Primary peer-reviewed study, Nature, 13 May 2026.
- Chinese Academy of Sciences report, 14 May 2026 — institutional account, not independent experimental replication.
- Nature Physics commentary, 15 June 2026 — separate technical discussion of the architecture and photon-loss problem.