What is the largest quantum computer by qubit count?

Much larger qubit counts circulate, and most of them are answering a different question. This page counts one thing: the largest disclosed qubit count on a system built and operated as a gate-based quantum computer, as a single physical device, backed by a linked source (paper, vendor announcement, or press coverage). Numbers that look bigger usually relax one of those conditions.

The common cases: roadmap or announced systems that have not been demonstrated as a working device; quantum annealers, which as built implement a different, non-universal computational model and cannot run gate-based algorithms like Shor's, however many qubits they report; fleet or cumulative counts summed across many machines; and modular architectures quoted at their interconnected design capacity rather than what has been operated as one system. None of those are wrong numbers, they are answers to different questions, and headlines rarely say which question was asked.

The inclusion rule cuts the other way too: a record here is a record demonstration, not a product you can buy or access, and a single-device record says nothing about yield, uptime, or gate quality at that scale. The line is genuinely fuzzy at the top: entries near the record may predate published gate benchmarks at full scale, so stricter criteria (entangling gates benchmarked on-device across the whole array) would shrink these numbers, not grow them. Every data point behind this page carries a source URL, and per-device details are available machine-readably through the site's REST API and MCP server.

Every record event in the dataset, across all tracked platforms, in chronological order. Qubit counts double roughly every 1.5 years (neutral atom), 1.4 years (superconducting), 2.3 years (trapped-ion) on log-linear fits of these series; the fits describe the record history and are not predictions.

Raw qubit counts overstate useful capacity. Useful fault-tolerant computation needs error-corrected logical qubits, each built from many physical qubits, see the overhead breakdown below. Dividing today's record count (1,180) by even the most efficient published code's overhead (~31.5 physical per logical for the Gross code) yields at most ~37 logical qubits, far short of the 1,399 needed for RSA-2048.

That division is a hosting-capacity estimate, not a demonstrated capability: it assumes every physical qubit on the record device could participate in the most efficient published code at sufficient fidelity, which no device has shown at this scale (the code itself is numerically established, not yet operated on hardware). It is the optimistic bound on what today's largest machine could mean in logical terms.