Q-CTRL has set three new performance records in magic-state cultivation using the publicly accessible IBM Quantum Nighthawk processor, surpassing previous demonstrations limited to Google’s Willow processor, the company says. The company’s performance-management software and novel protocols achieved a ten-fold reduction in state error, demonstrating a path toward more efficient factories for large-scale fault-tolerant quantum computing. The work achieves the lowest reported error rate for magic state cultivation, the lowest per-round logical error rate after escape in magic-state cultivation. The highest-fidelity magic state demonstrated in a surface code.
Lattice Surgery Demonstrates Record Magic State Teleportation
Q-CTRL achieved a surface-code fidelity exceeding 0.99 immediately following lattice surgery, establishing a new benchmark in magic state preparation and surpassing the previously reported record of approximately 0.91 for a magic state within a smaller surface code. This advancement centers on a technique called lattice surgery, where a cultivated quantum state is teleported directly into a standard surface code using established error-correction operations.
Unlike grafting, an alternative method, lattice surgery avoids intermediate code transformations that can accumulate errors and slow down decoding processes; the team’s results demonstrate the importance of evaluating a code transformation across its entire pathway, according to Q-CTRL. The experimental workflow involved injecting and cultivating a magic state within a small color code before employing either grafting or lattice surgery to transfer it into a distance-5 surface code, followed by postselection and decoding.
Executed on an IBM Quantum Nighthawk processor, Q-CTRL’s lattice-surgery protocol not only achieved record surface-code fidelity but also demonstrated a record memory fidelity of 0.991 after ten rounds of error correction, with a logical error rate of only 0.14 percent per round. This performance represents an improvement over the previous benchmark established on Google’s Willow processor, delivering roughly ten times lower error at a three times higher acceptance rate.
The success of lattice surgery is notable because it uses publicly accessible hardware. The IBM Quantum Nighthawk processor allows for broader impact and independent verification of these results. Proposed in 2025, this direct teleportation approach had not been demonstrated on physical hardware until now, highlighting a key step toward scalable quantum computation. The researchers explain that errors accumulate during intermediate steps and ultimately compromise the quality of the generated state.
Grafting, while ultimately reaching a decoder-friendly code, introduces a detour that sacrifices fidelity and necessitates slower decoding, whereas lattice surgery takes a more direct route, avoiding these overheads. The team’s manuscript details the first experimental demonstration of magic state cultivation with lattice-surgery escape, connecting cultivation, a method for preparing high-quality magic states, with the lattice-surgery operations needed to provide the state with greater quantum error correction protection, the company says.
This work establishes a new standard for generating the high-fidelity magic states essential for running universal algorithms on future fault-tolerant quantum computers, and demonstrates a pathway to reducing the costs associated with quantum error correction.
Magic State Factories & the Cost of Fault Tolerance
Lowering the overhead associated with generating high-fidelity magic states is critical for realizing practical, large-scale quantum computing, and recent work demonstrates a significant reduction in the resources required. This improvement stems from a novel approach combining magic state cultivation with lattice-surgery operations, offering a pathway to more efficient fault-tolerant quantum computation. The team’s protocol distinguishes itself from traditional magic state distillation by avoiding the need for extensive physical qubit arrays, instead employing a two-stage process of refinement and escape, Q-CTRL reports.
Refinement involves preparing a noisy logical magic state on a compact code and discarding any runs that detect errors, while escape transfers the verified state into a larger quantum error-correcting code for full fault tolerance. This cultivation method, first introduced in 2024, has now been experimentally demonstrated using publicly accessible IBM Quantum Nighthawk processors, allowing for broader verification of the results.
The researchers emphasize the potential for scalability. Beyond the initial fidelity boost, Q-CTRL’s experiments reveal sustained performance through multiple rounds of error correction. The combination of lower error and higher acceptance translates directly into fewer resources consumed during the magic state factory process, a key metric for assessing the viability of fault-tolerant quantum computers. The ability to achieve these results on publicly accessible hardware accelerates the pace of innovation and allows for independent validation of the findings, furthering the development of practical quantum computation.
Refinement and Escape Stages in Magic State Cultivation
This achievement demonstrates a direct transfer into a standard surface code, enabling syndrome data processing over 1,000 times faster than that required by grafted codes, a significant advantage for scalability. This initial stage is critical for reducing logical error before the state enters the more robust error-correcting phase, and the current work demonstrates an acceptance rate three times higher than previously observed on Google’s Willow processor.
Comparing the current work to their initial cultivation experiment on Google’s Willow processor, researchers found performance gains across most metrics used to evaluate fault-tolerant implementations, including a reduction in logical error from 1.9 × 10−4 to 2.0 × 10−5.




See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
