IBM Quantum has increased access to its fastest processor, the Nighthawk r2, offering 120 programmable qubits and a sample rate of up to 100,000 samples per second. The company highlighted the new hardware at the IBM Quantum Developer Conference 2026, emphasizing a shift toward developer experience as key to achieving quantum advantage. Developers can now run a 64-qubit experiment on Nighthawk r2 through an “onset of ergodicity” tutorial, and also gain direct pulse-level access enabling features like mid-circuit measurement and dynamic circuits.
Nighthawk r2 Processor Enables 64-Qubit Experimentation
Nighthawk r2 now facilitates 64-qubit experiments through an “onset of ergodicity” tutorial, allowing developers to directly assess the processor’s capabilities with a specific quantum workload. This demonstration demonstrates a significant increase in accessible qubit count for practical experimentation, moving beyond simulations to real hardware analysis. The tutorial provides a structured pathway for running and analyzing the experiment, offering insights into the behavior of quantum systems at this scale.
The processor’s speed, reaching up to 100,000 samples per second, is critical for exploring these frontier circuits. IBM’s device benchmarks are publicly available, enabling users to independently measure a processor’s maximum circuits per second (MCPS) and verify performance claims. This transparency allows for direct comparison between processors and encourages a community-driven approach to benchmarking.
For a more hands-on evaluation, the mirror benchmark runs a utility-scale circuit and checks its accuracy, forming part of IBM’s fully open benchmarking suite that also includes CLOPS, EPLG and more. Beyond simply running larger circuits, Nighthawk r2 provides developers with direct control over the quantum system through direct pulse-level access. This capability enables features like mid-circuit measurement, allowing for real-time feedback and adaptive quantum algorithms.
Fast qubit reset and dynamic circuit construction further enhance experimental flexibility, while soft (IQ) readout provides more precise measurement outcomes. This complements Qiskit Fermions by allowing developers to build circuits directly on bosonic modes without upfront encoding or truncation. The transpiler handles the boson-to-qubit mapping later, offering flexibility in choosing encodings like unary, binary and Gray.
This expanded access program aims to encourage more experimentation and facilitate longer-running workflows. The company also offers IBM Quantum Credits for IBM-funded access to larger workloads, IBM Quantum Classroom Accounts for educational purposes, and the Quantum Advantage Tracker, a community-driven platform for measuring progress toward quantum advantage.
Qiskit-Based Tooling for Quantum Workload Development
New usage estimation tools now provide IBM Quantum Platform users with greater visibility into resource consumption before execution, enabling more informed planning and optimization of quantum workloads. This pre-execution analysis allows developers to accurately assess the capacity required for their experiments, supporting confidence in resource allocation and experimentation. The platform’s expansion supports a broader range of quantum and classical computing capabilities, facilitating more sophisticated workflows and integration with high-performance computing environments.
IBM Quantum is introducing Resource Units (RUs) as a unified capacity model, where 1 RU equates to one minute of QPU compute time, ensuring a smooth transition for existing users and pricing structures. Beyond capacity planning, the platform now allows payment for classical compute resources using Resource Units, offering per-second billing across several classical compute profiles. This flexibility enables developers to deploy quantum-classical workflows beyond the limitations of local computing power, scaling up experiments with powerful remote environments.
Functionality like -powered dynamics simulation allows modeling of time evolution, specifying physical models, initial states and observable parameters to retrieve full time traces of expectation values. Catalog Functions are available on a consumption-based model, billed per-workload using resource units within Premium and Flex plans, removing the need for upfront commitments. Generative AI is accelerating quantum development, but general-purpose models often rely on outdated Qiskit APIs.
To address this, IBM is emphasizing Qiskit agent skills, open-source instructions that guide AI assistants to work with Qiskit as intended. Each skill is grounded in a curated capability registry and reference pages, ensuring the AI tools use current and supported functionalities. A particular focus is on application tooling that translates real-world problems into quantum circuits. Qiskit Fermions, for example, provides a toolkit for directly working with fermionic systems, relevant to chemistry, materials science and physics, before converting them to qubits.
IBM’s MCPS Benchmark Measures Nighthawk’s Circuit Performance
The Nighthawk r2 processor’s performance is now quantifiable through a new metric, circuits per second, allowing developers to directly assess its speed using the publicly available MCPS benchmark. This benchmark, part of IBM’s fully open suite alongside CLOPS and EPLG, provides a standardized method for evaluating processor capabilities and comparing results across different quantum hardware. The MCPS benchmark isn’t merely a theoretical test. Developers can immediately implement it using the qiskit_device_benchmarking library and a QiskitRuntimeService connection to the “ibm_nighthawk” backend.
Beyond the MCPS benchmark, a “mirror benchmark” offers a hands-on evaluation of circuit accuracy using a utility-scale circuit, further enhancing the transparency of performance testing. This directed execution model allows for composable transformations, such as gate and measurement twirling, to be applied before running a program as a single, repeatable workflow. This approach moves beyond isolated demonstrations, supporting larger research initiatives and the development of production-ready quantum applications.
The ability to package and run circuits with annotations enables reproducibility and scalability without sacrificing visibility into the underlying execution. The tools extend beyond simple execution, offering profiling and performance analysis features to optimize experiments before and after committing quantum processing unit time. Developers can now characterize qubits, visualize circuit scheduling, estimate resource usage and review the precise execution order of their circuits. Device drift, scheduling conflicts, noise and job timing all influence the quality of results.
Getting reliable outcomes, IBM argues, requires more than just a well-designed circuit. It demands a comprehensive understanding of the hardware and its behavior. The fermion-to-qubit mapping occurs during transpilation, simplifying the process of building complex simulations. The combination of these tools and the MCPS benchmark provides developers with a powerful suite for building, testing, and optimizing quantum algorithms, ultimately accelerating the path toward practical quantum computing.
Qiskit Paulice Framework Supports Error Correction Research
The Qiskit Paulice framework now supports the development of stabilizer-based computations and quantum error-correction workflows, offering developers a pathway to build with the primitives of fault-tolerant quantum computing. Installation is straightforward via pip: pip install qiskit-paulice, providing immediate access to the tools needed to begin experimentation. Developers can define a Clifford circuit, then use Paulice to identify optimal qubit pairings for error checks based on the processor’s connectivity map.
This process involves constructing a depolarizing noise model from backend data, allowing for a weighted scoring of potential checks to maximize fidelity. The resulting “checked circuit” incorporates these checks, enabling post-selection of error-free samples and a direct comparison of performance with and without error correction. This approach moves beyond theoretical exploration, offering a practical means of evaluating error-correction strategies on real quantum hardware.
The documentation details how to run the checked circuit, post-select samples and compare fidelities, providing a clear path for researchers to assess the effectiveness of different error-correction techniques. The release of Qiskit Paulice is part of a broader effort to provide developers with a comprehensive error-correction stack. Complementing the framework are modular error-mitigation tools, including qiskit-mitigation and qiskit-noise-learning, offering a layered approach to tackling quantum decoherence.
IBM emphasizes that this integrated suite spans the full stack, enabling developers to address error correction at multiple levels. Beyond the technical capabilities, the company highlights the accessibility of these tools. “Leverage the latest research software tools from our ecosystem to accelerate applied research iterations, now without procurement hurdles,” the company states, highlighting its commitment to open access and collaborative development.
Pulse-Level Access Facilitates Real-Time Control
IBM Quantum’s Nighthawk r2 processor now offers developers direct pulse-level access, a capability enabling features such as mid-circuit measurement, fast qubit reset, dynamic circuits, and soft (IQ) readout, fundamentally altering the granularity of control over quantum operations. This level of access allows researchers to manipulate qubits in real-time based on measurement outcomes, an essential step for advanced error-correction techniques and complex quantum algorithms. The ability to perform mid-circuit measurements and branch based on those results is particularly significant, forming the basis of syndrome extraction and feedback loops essential for error mitigation.
Researchers can now implement low-overhead instructions like MidCircuitMeasure to observe qubit states during a computation and adjust subsequent operations accordingly. This is demonstrated in a repetition-code experiment, now runnable end-to-end on IBM Quantum Platform, using experimental qec_experimental calibration alongside the MidCircuitMeasure instruction.
A tutorial on dynamic circuits, dc-hex-ising, further guides developers through custom dynamical decoupling and advanced scheduling techniques. Beyond real-time control, precise timing control is also now available, detailed in a new scheduling guide, alongside documentation addressing drift during long computational runs. This allows for optimized planning and execution of quantum workloads, and the ability to right-size experiments against resource allocations.
Users can perform a dry run to estimate a workload’s usage without consuming any quantum processing unit time, using the QiskitRuntimeService and Executor tools. The combination of high-performance hardware and accessible software tools positions IBM Quantum as a platform for pushing the boundaries of quantum computation and error correction.
Repetition Code Experiment Validates Long-Duration QEC Workflows
A repetition-code experiment involving one million mid-circuit measurement and reset layers is now executable end-to-end on IBM Quantum Platform, validating workflows essential for long-duration quantum error correction. This capability relies on the newly implemented for_loop instruction, which repeats operations without expanding the circuit’s size, a technique essential for scaling error correction routines.
The ability to run this experiment allows researchers to develop and test these workflows on real hardware, anticipating the demands of computations requiring millions of error-correction rounds. Beyond MCPS, the platform supports a broader benchmarking suite including CLOPS and EPLG, offering a comprehensive evaluation of processor capabilities.
“Know how much capacity your workload is expected to consume before execution, so you can optimize resource usage and experiment with confidence,” the company states. This allows for right-sizing experiments and verifying resource allocation before submission, streamlining the research process. This integration allows quantum computing steps to fit smoothly into established HPC workflows, eliminating the need for vendor-specific code. This expanded access and refined control are designed to support the development of increasingly sophisticated quantum workflows and error-correction techniques.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
