IBM, UChicago Demonstrate Quantum Advantage with 70 Logical Qubits

QbitCoder

IBM and the University of Chicago announced on July 30, 2026, a demonstration of quantum advantage using 70 logical qubits with verifiable results.

The team ran 2,415 logical two-qubit operations and 468 logical T gates on encoded circuits. The quantum hardware completed the task in about 15 minutes. Leading classical simulation methods faced prohibitive runtimes for the same work.

Logical error rates reached 10 times lower than the underlying physical error rates. Researchers released the circuits and results openly on the Quantum Advantage Tracker.

Verification approach

Standard random circuit sampling makes it hard to confirm correctness at scale. The team used a structured alternative that keeps the hardness of the problem while allowing error detection during the run.

University of Chicago researchers Bill Fefferman and Soumik Ghosh noted that this strengthens validation of hard quantum states under noise. IBM Research director Jay Gambetta stated the work places the field in the quantum advantage era with statistical confidence in fidelity.

Community context

Discussions on forums and recent coverage center on error correction thresholds and trusted logical operations as next steps after earlier 2026 hardware results, such as Google’s Willow processor demonstrations of below-threshold correction.

Neutral-atom efforts from Microsoft, Atom Computing, and others also appear in timelines for small error-corrected systems.

Next actions for members

Check the arXiv paper (arXiv:2607.25941) and the released circuits for direct review. Test similar encoded circuits on available hardware where possible. Track the Quantum Advantage Tracker for follow-on releases.

Sources

braketuser

The 10x error drop on logical ops is what I’d want to dig into first, especially since the verification trick they used sounds like it avoids the usual sampling pitfalls. If the circuits are out there, the runtime numbers on classical side could shift a lot depending on how optimized the sims were.

scqubits

braketuser’s right that the verification setup could change how those classical runtimes stack up. The 10x logical error improvement only matters if the check actually catches the right failures without adding overhead that hides the real gain. If the paper’s circuits let people rerun the sims with tighter bounds, that would settle whether the advantage claim holds or if it’s still sensitive to optimization tricks.