Every quantum computing company reports its qubit count and its own chosen benchmark, and until now there was no test that let you put Google's chip next to IBM's next to Quantinuum's and see who was actually further along. QPU-BM, a new benchmark reported September 15, ran the same suite of problems on hardware from all three companies and scored them against classical simulation on ordinary computers. The classical machines won by a factor of 100,000, on every platform, no exceptions.

That number matters because it is not about who has the most qubits. It is about logical qubits, the error-corrected unit that actually holds a calculation together long enough to be useful, as opposed to the raw physical qubits that companies put in headlines (a point the U.S. Department of Energy underscored in a separate September 17 assessment of the field). A processor with a thousand noisy physical qubits might yield only a handful of logical ones once you spend most of them on redundancy just to catch and correct errors before they corrupt the answer. QPU-BM tests what survives that overhead, and on that measure the gap between quantum and classical is not closing, it is just being measured honestly for the first time. The benchmark is a published methodology, not a chip; the next gate is whether any of the three companies posts a QPU-BM score that starts to close the 100,000x gap rather than just a bigger qubit count.