Doped Clifford Sampling Pairs Hardness Guarantees With an On-Chip Fidelity Certificate

Quantum advantage claims have long struggled with a verification problem: the hardest circuits are precisely those classical machines cannot check. In late July 2026, IBM and University of Chicago researchers published a structured alternative designed to combine hardness guarantees with an experimental fidelity certificate.

Doped Clifford sampling

In arXiv:2607.25941, collaborators from IBM Research and the University of Chicago describe doped Clifford sampling — circuits with provable hardness properties that can also be encoded in a quantum error-detecting code. The construction allows simultaneous high-depth execution and certification via code syndromes rather than full classical simulation.

The flagship experiment used a 70-qubit, depth-70 Clifford circuit doped with 468 T gates, encoded with spacetime codes across 97 physical qubits. After syndrome post-selection, the team reported a fidelity lower bound of 0.284 at 95% confidence.

Why IBM framed it as trusted advantage

IBM's July 30, 2026 blog post positioned the result as part of a broader trusted quantum computation push — three papers, including work with Qedma/RIKEN and Algorithmiq, aimed at regimes where leading classical methods become impractical or disagree with one another.

Unlike Google's contested 2019 random-circuit sampling narrative, the UChicago paper emphasizes structured hardness plus in-experiment certification. IBM added the doped-Clifford entry to its Quantum Advantage Tracker, where outside groups can attempt improved classical challengers.

What is — and is not — claimed

The preprint is explicit that the fidelity certificate is device-dependent and relies on weaker noise assumptions than some proxy benchmarks. IBM also treats tracker entries as active candidates, not adjudicated proofs — peer review is still pending.

That nuance matters: the work advances verification for sampling-style demonstrations, but the community still has to stress-test whether improved classical algorithms can close the gap.

Community benchmarking next

IBM reiterated that advantage claims live or die in public replication. The tracker already hosts candidates from Q-CTRL, BlueQubit, and others; the doped-Clifford challenge was exposed on the tracker weeks before the July 30 announcement campaign.

For quantum watchers, the headline is verification engineering: advantage demonstrations that ship with certificates, not just peak qubit counts.

Sources

• IBM Quantum — Researchers demonstrate quantum advantage through trusted quantum computation (July 30, 2026)
• arXiv — Sampling hard circuits with verifiably high fidelity (July 28, 2026)