Origin Quantum Says It Built a 180-Qubit Quantum Computer. Here Is What the Specs Actually Say.

Origin Quantum says it has built the most powerful superconducting quantum computer in China. The state media version of that sentence, published this week, added the words "breakthrough," "fourth generation," and "globally leading." The company's own blog, posted the same day, said something more modest: the new Wukong-180 has 180 qubits, which is 2.5 times more than its predecessor, and it works.

Both things can be true. The gap between them is the story.

Origin Quantum Computing Technology, based in Hefei, launched Wukong-180 publicly on May 9th, describing it as the first time a Chinese company has made a domestically developed superconducting quantum computer with more than 100 qubits available to users worldwide. The system runs on a single chip with 180 computational qubits and 251 coupling qubits. The company is offering free compute time through its cloud platform — a genuine commercial first for OriginQC, which previously operated China-only.

The specs OriginQC published on its blog: average T1 coherence time of 40 microseconds, T2 echo coherence of 20 microseconds, single-qubit gate fidelity of 99.9 percent, two-qubit gate fidelity of 99 percent, and readout fidelity of 99 percent. Those are self-reported numbers from a company with an IPO filing pending and every incentive to publish clean data. Nobody outside OriginQC has published an independent benchmark of Wukong-180 yet.

That is the part the state media framing quietly skipped.

The arXiv preprint that comes closest to independent confirmation is a paper from April 2026, in which researchers at the National Supercomputing Center in Zhengzhou and Zhengzhou University used Wukong-180 to run experiments validating a quantum error mitigation technique. The circuits involved 10 to 16 qubits — not 180, not even 50. The paper confirms the machine is operational and accessible. It does not benchmark Wukong-180 against anything IBM or Google has published.

Which raises the question: how does 180 qubits on a Chinese superconducting chip compare to what the leading Western players have?

IBM's current flagship superconducting system, Heron r3, reports T1 coherence times of 300 to 500 microseconds — roughly 10 times longer than Wukong-180. IBM's median two-qubit gate error rate is around 2 × 10⁻³. OriginQC's self-reported two-qubit gate fidelity of 99 percent implies an error rate of roughly 1 × 10⁻². That is roughly five times worse than IBM's current publicly verified number. The comparison is rough — OriginQC's blog doesn't break out error rates the same way IBM does in its published calibration data — but the direction is clear: Wukong-180 is a meaningful step up from OriginQC's own prior systems, and a notable gap behind IBM's best.

The "fourth generation" label OriginQC uses for Wukong is a product iteration name, not a physics threshold. There is no international standard for what constitutes a quantum computer generation; the term reflects OriginQC's own roadmap versioning. State media picked it up and turned it into a breakthrough claim. The company's own technical blog is more careful: it calls the machine a "tangible leap in scalable quantum processing" and describes the 180-qubit count as a crossing of an engineering threshold — "the 100-qubit engineering phase ready for real-world delivery." That's real language, not hype language.

What OriginQC appears to have actually done is scale qubit count substantially while holding coherence and gate fidelity roughly flat relative to their prior system. Whether that tradeoff is the right engineering bet depends on what workloads the machine is asked to run. The company is positioning Wukong-180 for quantum chemistry, optimization, and AI applications — markets where qubit count matters until it doesn't, and where error rates eventually dominate everything else.

The honest version of this story: a Chinese quantum computing company with a pending IPO has shipped a larger machine with self-reported specs, no independent benchmarks yet, and a state media amplification operation that turned a product launch into a geopolitical statement. The machine may work exactly as OriginQC describes. The gap between what the company says and what has been independently verified is not trivial.

For buyers evaluating quantum hardware through actual cloud access rather than press releases, the relevant number isn't the qubit count. It is whether the error rates on their specific workload are low enough to produce useful answers. Wukong-180's self-reported two-qubit error rate of roughly 1 percent would require error mitigation or correction to approach fault-tolerant territory — a constraint IBM also faces at similar scales, but one that IBM has published significantly better numbers to partially address.

OriginQC says the system is available globally. Whether that means a Western researcher can log in and run circuits today is a separate question worth asking before acting on any of this.