IBM claims quantum advantage with three new error-mitigated algorithms

IBM announced three quantum advantage demonstrations on July 30, 2026, each tackling errors and verification with a distinct approach.

The work addresses a persistent obstacle: proving a quantum computer’s superiority for tasks beyond classical reach while ensuring the output can be trusted. “Trusted computing when you can’t do classical simulations is a big deal,” said Jay Gambetta of IBM.

Past quantum advantage claims have been undercut when improved classical algorithms caught up or systemic errors went undetected.

In one effort, IBM, RIKEN, and Qedma modeled a Floquet Ising model on an IBM processor with Qedma’s error-mitigation software. Classical algorithms on the Fugaku supercomputer clashed, whereas the quantum output revealed a gradual magnetism decline with periodic oscillations—a pattern later verified on a Quantinuum device.

Another demonstration, from IBM and the University of Chicago, used a sampling algorithm built mainly from Clifford gates and a handful of T gates. The T gates are less noisy and make classical simulation exponentially difficult. Peripheral qubits spotted errors, discarding faulty shots.

A third result came from Algorithmiq, which ran an echo-based algorithm on a low-noise patch of an IBM chip. Neighboring qubits flagged faults, and intentional noise injection let the team bound the error rate.

IBM launched a quantum advantage tracker to follow such claims as classical methods evolve. Previous entries have fallen, but the new results emphasize rigorous quality over qubit-count milestones.

The field now seeks to leverage these techniques for algorithms with real-world impact. “The holy grail is to compare a real material or a real experiment to a quantum computer,” Gambetta said.