AIBRIEF
IBM's three arXiv preprints include one that claims its 156 qubit Heron R3 quantum processor did in minutes what Japan's Fugaku supercomputer would need years to do.
IBM, with partners at Algorithmiq, Qedma, and the University of Chicago, posted three arXiv preprints on July 27 and framed them at a July 28 news conference as evidence of quantum advantage on the company's 156-qubit Quantum Heron R3 superconducting processor.
The first preprint, on simulating a Floquet transverse-field Ising model with error mitigation from Qedma, is the one IBM pairs with a minutes-versus-years claim against Japan's Fugaku supercomputer. The other two cover observable estimation without classical verification and a heuristic peaked-circuit advantage. All three are preprints, not peer-reviewed.
IBM is leaning into the open audit, not the boast. The company has invited outside challenge through the community-run Quantum Advantage Tracker, a GitHub ledger built to test claims like these. Speed numbers rot; the Tracker is where a sustained rebuttal would land.
Three narrow, problem-specific results on one processor, with peer review and independent replication still pending. Watch the Tracker for the verdict.