A proposed verdict (marked not yet final) and the final verdict, for every claim on the Ledger. Nothing else. https://symbiq.dev/feeds/verdicts.xml 2026-09-22T00:00:00Z SymbiQ deskhttps://symbiq.dev/ tools/build_verdict_feeds.py urn:symbiq:ledger:ionq-five-nines-2025:resolved:2026-09-22:0 2026-09-22T00:00:00Z 2026-09-22T00:00:00Z

IonQ's best published result is about 99.9916% on physical gates with no error correction; the claim was 99.999% on logical gates.

Evidence gathered from the pre-registered resolver sources only. The tracked target, as stated in IonQ's July 2024 roadmap announcement, was 99.999% (“five nines”) LOGICAL two-qubit gate fidelity using small-distance quantum error-correcting codes, by the end of 2025. Inside the window (31 Dec 2025 plus the 30-day grace, i.e. by 30 Jan 2026) IonQ did publish a two-qubit gate result: on 20–21 October 2025 it reported a gate error of 8.4(7)×10⁻⁵ — about 99.9916% fidelity — using Electronic Qubit Control without ground-state cooling, demonstrated on prototype systems in IonQ's R&D labs (arXiv:2510.17286; IonQ newsroom; The Quantum Insider, 21 Oct 2025). That is a PHYSICAL (native) gate fidelity. No error-correcting code is involved, and it is measured on prototype hardware rather than a deployed system. It is a real advance, it was reported as a world record on the physical axis, and it comfortably clears the same roadmap's separate >99.9% native target — but it is not the claim tracked here. No result at or above 99.999% logical two-qubit gate fidelity with small-distance QEC codes appears in any named resolver source inside the window. On the PARTIALLY_VERIFIED branch, recorded so the reading is auditable rather than assumed: the criteria offer that branch for “a result close to but short of five nines (e.g. four nines) … reported as the headline achievement instead.” The desk reads it as scoped to LOGICAL fidelity, because the sentence immediately before it forbids conflating the native figure with the logical one. A four-nines PHYSICAL result is a different measurement, not a near miss on the same one, so the branch does not fire. This is a statement about a dated claim and its deadline, not about the people who made it. urn:symbiq:ledger:atom-computing-microsoft-24-logical-qubits-2025:resolved:2026-09-22:0 2026-09-22T00:00:00Z 2026-09-22T00:00:00Z

No customer was operating a 24-logical-qubit system by 30 January 2026; the one commercial sale, QuNorth's Magne, is expected to be operational in spring 2027.

Evidence gathered from the pre-registered resolver sources only. The criteria required a commercial customer to have RECEIVED AND OPERATED a system matching the 24-entangled-logical-qubit specification by 31 Dec 2025 plus the 30-day grace (30 Jan 2026), and said in advance that an order or a demo does not count. One commercial sale is identifiable in the named sources: QuNorth (Denmark, funded by EIFO and the Novo Nordisk Foundation), which contracted in 2025 for Magne — an Atom Computing / Microsoft system specified at roughly 50 logical qubits on about 1,225 physical neutral atoms, announced July 2025 with construction beginning that autumn. As of 2 September 2026, more than seven months after the grace period closed, The Quantum Insider reports that the building intended to house Magne has only just been completed, that installation is beginning, and that the system is expected to be fully operational in spring 2027. So no system was operating at a customer site inside the window — not at the announced 24-logical specification, and not at the differently-specified 50-logical one. That also rules out PARTIALLY_VERIFIED, which required a differently-specified system to have shipped in the window; none did. The January 2025 HPCwire roll-out report logged at intake describes an offering being made available, not a delivery — exactly the distinction the criteria fixed in advance, which is the whole reason it was fixed in advance. Nothing here is a judgement about intent. Hardware schedules slip, and the original announcement was explicit that delivery was a 2025 plan. What the Ledger records is that the dated, falsifiable version of the claim did not come true inside its own window.