Chinese Team Solves Quantum Benchmark on Supercomputers
Researchers used exascale classical machines to compute a quantum chemistry benchmark's ground-state energy in 105 minutes.
Updated on Nov. 20, 2026 in Quantum Computing

Chinese researchers from seven institutions used exascale supercomputers to solve the FeMoco ground-state energy benchmark in 105 minutes, a result reported at the SC26 conference in Chicago in early November 2026. The work was done on classical machines rather than quantum computers.
Why it matters
FeMoco's ground-state energy has served as a benchmark for quantum computing performance, and the classical solve challenges claims about the time a fault-tolerant quantum machine would need. The result shows how far classical high-performance computing can now reach on problems once expected to require quantum hardware.
The FeMoco model contains 113 electrons across 76 orbitals, and the code ran at 1.456 EFLOP/s in double precision and 5.626 EFLOP/s in mixed precision across 20,480 nodes of the LineShine and CNIS systems. The source does not indicate whether these performance figures will hold for other chemical systems.
The players
SC26 conference
SC26 is the annual international high-performance computing conference where the result was presented in Chicago in November 2026.
The details
Researchers used the LineShine exascale supercomputer and China's CNIS to compute the lowest-energy ground state of the iron-molybdenum cofactor, after benchmarking the code across thousands of nodes. Earlier in 2026, other researchers had shown that classical methods could also estimate the FeMoco energy.
Timeline
2019: The 113-electron, 76-orbital FeMoco model was defined as a benchmark.
Earlier in 2026: Other researchers demonstrated classical estimation methods for FeMoco's ground-state energy.
Early November 2026: The 105-minute solve was reported at the SC26 conference in Chicago.
Quantum Computing Benchmarks
The FeMoco model was defined as a quantum computing benchmark in 2019, and the 2026 classical solve in 105 minutes now sets a competing reference for how long that problem takes. The gap between that 105-minute classical solve and the 8.5-hour fault-tolerant quantum estimate will shape how future benchmarks are judged.
For most people the result does not change anything they use directly today. It does affect how researchers judge when quantum computers will outperform classical machines on chemistry problems.
The takeaway
Classical supercomputers can now solve a problem once expected to need quantum hardware, so benchmarks for quantum advantage will need to be revisited as classical methods improve.
Further reading
For background on the quantum systems this classical result is measured against, see Quantum Computing.
Source note: This article includes information reported by Wccftech.



