Overview
- The research teams announced coordinated demonstrations on Thursday showing quantum computers completed calculations that leading classical simulators could not reliably reproduce while offering built-in checks to increase confidence in the outputs.
- IBM and the University of Chicago ran an encoded logical computation using 70 logical qubits that executed 2,415 logical two-qubit gates and 468 logical T gates, cut logical error rates roughly tenfold, and finished a task in about 15 minutes.
- IBM and Qedma used QESEM error-mitigation software to track long-lived quantum dynamics up to roughly 74 qubits where multiple state-of-the-art classical methods, including runs on Fugaku, produced inconsistent results.
- IBM and Algorithmiq reported a heterogeneous-matter simulation first posted to the Quantum Advantage Tracker eight months ago that has resisted full classical reproduction and released a classical benchmark named monoprop for public testing.
- All teams released circuits, raw results, and preprints and used cross-device validation, deliberate noise manipulation, and open benchmarking so the community can attempt independent classical refutations while noting these advances do not change near-term cryptographic risk.