IBM announced the availability of its IBM Quantum Nighthawk r2 quantum processor through the IBM Quantum Platform, in a move focused on increasing the amount of useful computing that can be performed during a single runtime rather than simply increasing the number of qubits. The system has 120 programmable qubits, and IBM says it can execute more than 100,000 circuits per second—up to 25 times the rate of Heron processors, which is approximately 4,000 circuits per second.
The company attributes this improvement to a new architecture for resetting qubits quickly and independently. According to IBM, Nighthawk r2 also produced accurate results on circuits containing more than 7,500 gates, an achievement the company lists among the goals of its 2026 roadmap.
Removing the Bottleneck Between Circuits
A quantum circuit needs the qubits to begin in a known state before execution and to be returned to the ground state after measurement and before the next circuit runs. In earlier generations, including Heron, IBM relied on conditional reset, which measures the qubit and then applies a pulse to flip it if it is in the |1⟩ state. However, this method is affected by measurement accuracy and does not address qubits that leak outside the computational state, which may impose a waiting period of up to hundreds of microseconds.
Nighthawk r2 replaces this with what IBM calls the dissipative reset gadget. Each programmable qubit is connected to a cold environment through a tunable coupler with a high dynamic range. When the coupler is activated, the effective relaxation time, T1, drops from an average of approximately 200 microseconds to about 25 nanoseconds. The company says the idle time between circuit executions can be reduced to one microsecond, while neighboring qubits remain undisturbed.
The benefit is particularly apparent in large-scale, repeated workloads. IBM pointed to initial tests on circuits designed to demonstrate quantum advantage that achieved speedups of up to 10× without loss of accuracy. According to the company, the system also enabled a 12× speedup in neutron-scattering simulations and produced spectra directly comparable with laboratory data in approximately 60 seconds.
More Components Than the Advertised Qubit Count
The processor does not consist solely of programmable qubits. In addition to 120 qubits, Nighthawk r2 contains 218 dedicated couplers and 120 independent reset elements, bringing the total to 458 actual quantum components. IBM describes the system as the most complex quantum processor it has produced to date because the additional support elements require controllability and manufacturing capabilities no less important than those of the qubits that users program directly.
IBM says the processor maintains gate fidelity at the Heron level while reducing initialization error by approximately 25×. The reset mechanism was also designed to be safe for neighboring qubits, which is important in the square-grid architecture, where most qubits are connected to four neighbors, compared with two or three in earlier architectures.
Direct Significance for Error-Correction Research
The new reset mechanism operates not only between circuits but also during circuit execution. This makes it suitable for dynamic circuits that include qubit measurements during execution, a mechanism used in many protocols for detecting and correcting errors and in workflows that combine quantum and classical computing.
IBM notes that the ability to reuse auxiliary qubits repeatedly supports experiments such as space-time checks, in addition to research involving logical qubits. Researchers also used Nighthawk r2 in experiments involving interleaved Clifford sampling associated with demonstrations of reliable quantum computing, and in estimating observables using the Probabilistic Error Amplification technique on circuits exceeding 7,500 gates.
What Changes in Practice?
The main improvement here is not solely the qubit count, but the reduction of unproductive time between executions while maintaining the level of quality that IBM compares with Heron processors. This could give researchers a greater number of experiments within the same allocation and reduce the time consumed by reinitialization in repeated workloads and dynamic circuits.
Nevertheless, the performance figures and experimental results presented are based on IBM’s own announcement, and the source does not provide an independent comparison or complete details about the measurement conditions and the limits of system availability on the platform. Nighthawk r2 therefore represents an important advance in the development of quantum hardware, but it does not by itself demonstrate the achievement of fault-tolerant quantum computing at a practical scale; rather, it provides a faster experimental platform for studying that path.