IBM and the University of Chicago report a „verified quantum advantage”: a quantum computer is said to have solved a specially constructed task in about 15 minutes that, according to the researchers, would take leading classical simulation methods an impractically long time – running on 70 error-corrected logical qubits.
What the team did
According to IBM, the machine carried out 2,415 logical two-qubit operations and 468 so-called T gates, with logical error rates roughly ten times lower than the physical ones. Each logical qubit is encoded from a much larger number of physical qubits to catch computational errors. The work is described in the preprint „Sampling hard circuits with verifiably high fidelity” (arXiv), which has not yet been peer reviewed. Instead of classic random circuit sampling, the team says it uses a structured variant whose result can additionally be checked statistically.
Why experts remain cautious
The demonstration concerns a purpose-built sampling problem with no immediate practical use. Here „not classically reproducible” specifically means that known simulation methods hit runtime limits – it is not a mathematical proof of impossibility. Earlier quantum-advantage claims have repeatedly been matched later by improved classical algorithms. UChicago researcher Bill Fefferman calls verification itself „one of the biggest challenges” in establishing experimental quantum advantage, while IBM research director Jay Gambetta states the field is „now firmly in the quantum advantage era.”
Context
Nothing changes for users in the short term: the milestone shows progress in error correction and verifiability, not a ready-to-use quantum algorithm. Because the results so far exist only as a preprint, a peer-reviewed publication and independent replication by other research groups are still pending.
Sources: IBM Newsroom · arXiv preprint 2607.25941 · ScienceDaily · The Quantum Insider
