IBM Demonstrates Verified Quantum Results That Challenge Classical Methods

IBM together with researchers from the University of Chicago, Qedma, and Algorithmiq announced on July 30, 2026, three coordinated demonstrations of quantum advantage. The experiments, executed on IBM Quantum Heron processors, perform computations that challenge current leading classical simulation methods while introducing methods to build confidence in results when exact classical verification is unavailable.

Logical Encoding and Floquet Dynamics

In the collaboration with the University of Chicago, researchers constructed encoded quantum circuits that retain the hardness of random circuit sampling while enabling error detection.

Using spacetime codes, the team executed encoded circuits representing up to 70 logical qubits, incorporating 2,415 logical two-qubit operations and 468 logical T gates. The logical encoding achieved up to ~10× lower effective error rates in tested circuits relative to underlying physical error rates. The quantum system completed the sampling task in approximately 15 minutes, a duration that leading classical simulation approaches appear computationally prohibitive for known classical methods in this regime.

Validation integrated syndrome checks to derive a statistically rigorous lower bound on the fidelity of the encoded computation.

In parallel, Qedma applied its QESEM (Quantum Error Suppression and Error Mitigation) software to IBM Heron processors to model the long-time dynamics of a two-dimensional Floquet Ising system. Circuits of up to 74 qubits resolved persistent oscillatory behavior in magnetization under periodic driving.

Classical tensor-network and related methods, executed on RIKEN‘s Fugaku supercomputer in partnership with BlueQubit, produced diverging or inconclusive predictions at larger scales. Independent runs on Quantinuum trapped-ion hardware reproduced the same quantum oscillatory signature, increasing confidence that the results reflect the underlying physical model rather than device-specific effects.

Process Validation and Open Benchmarking

Algorithmiq demonstrated a quantum simulation of heterogeneous quantum matter by estimating the operator Loschmidt echo on IBM Heron systems at scales around 56 qubits. In the studied regime, multiple independent classical simulation groups showed increasing disagreement across classical methods at scale.

Trust was established by executing the algorithm across five IBM quantum processors under controlled noise injection and varied calibrations; the quantum outputs remained stable. Algorithmiq further released monoprop, an open-source classical simulation package, to facilitate community stress-testing of advantage claims.

The instance has remained competitive on the Quantum Advantage Tracker for eight months without a confirmed classical match across the full regime.

QESEM is available to users through the IBM Quantum Platform and the Qiskit Functions Catalog, providing commercially available error mitigation tools on current hardware. Jay Gambetta, Director of IBM Research and IBM Fellow, stated that the results indicate progress toward quantum computers that can outperform classical methods in select regimes and produce solutions that can be trusted through extensive testing.

Circuits, data, and validation protocols from all three demonstrations have been released to the Quantum Advantage Tracker to support continued independent scrutiny.

Bottom Line

IBM and partners have demonstrated evidence of trusted quantum computations that challenge current classical simulation methods on commercially available hardware.