IBM’s Nighthawk r2 Cuts the Wait Between Quantum Circuits
A new reset mechanism lifts IBM’s reported maximum circuit throughput above 100,000 per second, with gains that still depend on the workload.
IBM Quantum System One at the Thomas J. Watson Research Center in Yorktown Heights, New York, photographed in 2024. Context image for IBM quantum hardware—not Nighthawk r2 or a verified Nighthawk installation. Image credit“IBM Quantum System One.jpg” — Onri Jay Benally / OJB Quantum, CC BY 4.0, via Wikimedia Commons. Creative Commons Attribution 4.0 International. · https://creativecommons.org/licenses/by/4.0/
Quantum experiments often repeat a circuit thousands of times, making the wait between runs a real cost. IBM's Nighthawk r2 targets that pause with a dedicated reset element for each of its 120 programmable . In its August 31 announcement, IBM describes a controllable connection to a cold environment that draws each qubit toward its starting state. The company says qubits can reset independently without disturbing their neighbors, while retaining gate performance comparable to its Heron processors.
IBM's September service update lists a default repetition delay of one microsecond, down from 250, and maximum throughput above 100,000 circuits per second. The first r2 system, ibm_phoenix, is listed for Premium, Flex and Pay-As-You-Go users. IBM reports up to 25 times Heron's circuit throughput and, separately, early runtime gains of up to tenfold on some demanding circuits.
Those are different measurements. Faster resets do not accelerate every gate, data transfer or classical calculation by the same factor, and maximum circuit throughput does not determine an application's runtime. The useful test is how long a particular workload takes to reach the same answer quality, including repeated measurements, error handling and classical processing. IBM's company-reported results identify a concrete reduction in waiting time; they do not establish a universal speed advantage over classical computers.