Zuchongzhi and Jiuzhang: China's Two Quantum Computer Families Explained

Updated 2026-10-11 | 4 min read | QUANTUM (QNT) community

Two families, two kinds of qubit

USTC, led by Pan Jianwei and colleagues, builds two very different machines. Zuchongzhi uses superconducting circuits, the same family as IBM and Google chips. Jiuzhang uses light, in the style described in photonic quantum computing. Having both lets China learn from each approach. For the lab behind them, see the Hefei hub.

Zuchongzhi 3.0: 105 qubits

Search reporting on the peer-reviewed paper (Physical Review Letters, published March 2025) says Zuchongzhi 3.0 has 105 qubits and 182 couplers in a 15 by 7 array, with reported fidelities of 99.90 percent for single-qubit gates, 99.62 percent for two-qubit gates and 99.13 percent for readout, and a coherence time of 72 microseconds. The team ran an 83-qubit, 32-cycle random circuit sampling task and estimated that the Frontier supercomputer would need about 5.9 billion years to replicate it. They describe this as six orders of magnitude beyond Google's earlier 67 and 70 qubit experiments.

Here is the caveat the APS Physics commentary by Barry Sanders raises, as summarized in that reporting: the classical cost is not measured directly. Researchers infer it from reasonable assumptions about the best known simulation method, and those assumptions can be contested. Random circuit sampling is also a benchmark, not a useful application. For the wider debate, read quantum supremacy and advantage and quantum benchmarks.

Zuchongzhi 3.2: the error correction milestone

Arguably the more important result came in December 2025. A PostQuantum write-up reports that USTC ran a distance-7 surface code on a 107-qubit processor (97 qubits used for the code) and measured an error suppression factor of 1.40, published in Physical Review Letters on December 22, 2025. That crosses the below-threshold line, meaning adding more qubits made logical errors fall instead of rise. This is the idea behind quantum error correction, and it makes China the second group reported to reach this milestone after Google's Willow in December 2024.

The same write-up gives an honest comparison: Willow's suppression factor was 2.14, so Google's error suppression is stronger, Willow also showed real-time decoding, which USTC had not demonstrated. USTC used an all-microwave method for leakage suppression, which the author presents as potentially easier to scale. See Google Willow for the other side.

Jiuzhang 4.0: a photonic record

In May 2026 the team reported Jiuzhang 4.0 in Nature, according to a PostQuantum summary and a USTC news item. It is a Gaussian boson sampling machine with 1,024 squeezed-state inputs and 8,176 modes. The team reports detecting up to 3,050 photon clicks in a supplementary dataset (255 for Jiuzhang 3.0 in 2023), and estimates the El Capitan supercomputer would need over 10 to the 42 years per sample versus about 25.6 microseconds on the machine, a claimed speedup above 10 to the 54.

Now the strict honesty part, all from that same summary:

The 3,050 figure and the main runtime estimate come from different datasets (the main one maxes at 2,598 clicks).
Gaussian boson sampling is a specialized sampling task. The machine has no universal gate set, no error correction, and cannot run Shor's algorithm.
Classical simulation methods that exploit photon loss have challenged earlier photonic claims, so speedup estimates are contested in principle.
The team made raw data public so others can test the claims, which is a healthy sign.
The detectors are cryogenic, so the full system is not room temperature.

Commercial access

One secondary report says a system based on the Zuchongzhi 3.0 design has been deployed for commercial use through China Telecom Quantum Group, QuantumCTek and the Tianyan cloud platform. Treat that as less authoritative than the papers.

The optimistic takeaway

Whatever the exact speedup, the pattern is encouraging: more qubits, higher fidelities, error correction that works below threshold. Progress in two countries at once makes the whole path to useful machines more credible. Useful, fault-tolerant machines are still ahead, and experts do not expect public-key encryption to fall in the immediate future.

Sources and further reading

Reported as of 2026-10-09. Press and company claims change, so check the primary documents before relying on any figure. Nothing here is financial advice, and the QUANTUM (QNT) memecoin is independent and has no link to any lab, company or government mentioned.

Frequently asked questions

Did China prove quantum supremacy?

Chinese teams claim quantum advantage on specific sampling tasks. The classical cost is estimated, not directly measured, and the tasks are not useful applications, so experts treat the claims with care.

Is Zuchongzhi better than Google Willow?

Not on error suppression. Reported figures put Willow's suppression factor at 2.14 versus 1.40 for Zuchongzhi 3.2, though both crossed the below-threshold line.

Can Jiuzhang break encryption?

No. Gaussian boson sampling is a narrow task with no universal gates or error correction and cannot run Shor's algorithm.

Keep reading

Hefei National Laboratory and USTC: The Engine Room of China's Quantum Push
How the Hefei quantum hub, the University of Science and Technology of China and Pan Jianwei fit together, and what is and is not known about the money.
Quantum Error Correction in 2026: Where the Race Really Stands
A plain English scorecard of error correction progress: below-threshold results, logical qubit demonstrations, magic states, qLDPC codes and what is still unproven.
China vs the West in Quantum: An Evidence-Based Scorecard
Who leads where? A balanced look at computing, error correction, networking, talent and investment using the June 2026 SCSP scorecard and peer-reviewed results.
Photonic Quantum Computing Explained
How does photonic quantum computing work? See how single particles of light become qubits, why photon loss is the main hurdle, and where the approach stands.

All Quantum policy and governments guides | Back to top | Search the site

Main pages: Quantum computing explained | Quantum and crypto | Companies | Quantum news | Glossary

QUANTUM (QNT) is the quantum sector memecoin on Solana. See the live chart, buys and burnt supply or read the token facts. Questions? Join the Telegram.