QCMet: quantum computing metrics software package for full-stack benchmarking from low-level noise sources to applications

To verify the performance of quantum computers and guide their progress toward scalable implementations achieving quantum advantage, it is essential to evaluate performance metrics across all levels - from the individual sources of noise in qubits to full applications that can outperform classical computers. Here we present the QCMet software package that addresses this need [1]. QCMet implements a comprehensive collection of benchmarks allowing holistic comparisons of the performance of quantum computers, and is linked to a systematic and consistent set of definitions across all metrics, including a transparent description of the methodology and of the main assumptions and limitations [1]. We show how widely used performance metrics link to low-level sources of noise such as two-level systems in superconducting qubits, and present results for different hardware platforms. By making the software fully open source and easy to access it creates the basis for future developments of standardized approaches to quantum computing performance benchmarks, allowing to enable trust in the end user as well as to accelerate the progress of quantum computing hardware towards quantum advantage.

[1] D. Lall et al., A Review and Collection of Metrics and Benchmarks for Quantum Computers: Definitions, Methodologies and Software, arXiv:2502.06717; https://qcmet.npl.co.uk