In this article, we present the main ideas and results of a recent research work introducing the Many-body Quantum Score (MBQS), a benchmark designed to evaluate quantum computers on physically meaningful tasks. The complete paper details the theoretical model, benchmarking protocol, and experimental validation.
Quantum computing platforms are becoming more numerous and more advanced. To keep track of progress and to let users evaluate fairly which quantum computer (QC) is best for them, it is important to produce benchmarks. But can we actually compare these machines?
Benchmarking has long been a standard in classical computing. For example, supercomputers are ranked through the TOP500 to transparently measure and compare their performances. Similarly, in AI and deep learning, the 2012 ImageNet large scale visual recognition challenge (ILSVRC) challenged researchers to build algorithms capable of recognising objects in images at large scale. Ultimately, it provided a shared framework for evaluation, helping researchers measure progress in a consistent way.
For quantum computing to mature in a similar way, developing clear and widely accepted benchmarks is essential.
Benchmarks should respect many criteria.
Most importantly for quantum computers, they should be scalable and still be easy to run once quantum computing technology will be ready to reach hundreds or thousands of qubits!
They should also be related to performing a useful task, i.e. be application focused. Here, we focus on the task of solving quantum many-body problems, meaning systems where many quantum particles interact simultaneously and cannot be described independently due to effects such as quantum entanglement. These problems arise, for example, when studying superconducting materials or developing new molecules.
Finally, they should be relevant to compare different quantum computers architectures, such as digital and analogue quantum computers.
We developed, in collaboration with CEA, the French Alternative Energies and Atomic Energy Commission, the Many-Body QScore (MBQS), inspired by our previously introduced QScore.
It consists in reproducing the dynamics of a particular transverse field Ising model: the spinsIntrinsic quantum property of particles (such as electrons) that behaves like a two-level system and that are used to physically encode the 0 and 1 states of a qubit are placed on a circle. At the start, all spins are independent: they interact only with their nearest neighbours and are initially prepared in an uncorrelated (product) state, meaning that the state of one gives no information about the others.
As the system evolves, quantum correlations spread across the ring. They are arising from quantum entanglement progressively build up between spins that are increasingly far apart. This is quantified by measuring correlations, with a particular focus on spins located opposite each other on the circle.
Due to the particular nature of this Ising model, this build-up of long-range correlations is a well-understood effect that can be predicted theoretically. More specifically, the exact amount of correlation which should be obtained by a perfect quantum computer, is known for any number of spins.
In practice, real quantum computers achieve lower correlations due to noise and imperfections, making this a robust way to assess computational accuracy.
The MBQS is defined as the largest number of spins for which the measured long-range correlation reaches at least 50% of the ideal value (other percentage levels can also be considered).
The MBQS was evaluated on Ruby, an analogue quantum processor based on Rydberg atoms and developed by Pasqal. This initial implementation demonstrates the practicality of the approach, particularly for analogue quantum computing platforms. It also provides a first reference result against which other quantum computing architectures can be compared.
We hope this benchmark will be a useful tool for the community to track and evaluate progress in quantum computing, particularly in applications related to chemistry and materials science. Indeed, the ability to generate non-trivial long-range many-body correlations is both a fundamental challenge and a key requirement for enabling impactful real-world applications.
To go further, you can access our paper on Many-body Quantum Score: a scalable benchmark for digital and analog quantum processors and first test on a commercial neutral atom device, published on January 6, 2026.
We view this work as a reflection of a collaborative effort, illustrating what can be achieved by combining Bull expertise with our Qaptiva platform, which supports analytical theory, classical simulation, and noise modelling.
We also gratefully acknowledge the support of TGCCTGCC (Très Grand Centre de Calcul) is the supercomputer of the French Alternative Energies and Atomic Energy Commission (CEA), operated by the CEA at its Saclay research centre. Designed for large-scale numerical simulations, it is one of the cornerstones of France’s national high-performance computing infrastructure and supports high-performance scientific and industrial research. and the BACQ projectCoordinated by Thales Group and supported by a consortium of industrial and research stakeholders including the CEA, the BACQ project aims to define robust, meaningful, and user-friendly benchmarks to objectively assess the performance of next-generation quantum computing technologies..