Researchers at the University of the Basque Country and BCAM published QuSquare, a benchmark suite for evaluating quantum devices before fault tolerance. The paper appeared in Quantum Science and Technology (DOI 10.1088/2058-9565/ae917c) with an open-source implementation.
The problem the suite targets
Hardware vendors publish many different metrics, and results rarely compare across architectures. QuSquare's authors argue misleading performance numbers distort research priorities. The suite builds relevance, reproducibility, fairness, verifiability, and scalability into four tests for fair comparisons across superconducting, trapped-ion, neutral-atom, and photonic systems.
The four benchmarks
The suite runs a Partial Clifford randomized benchmark for gate accuracy, a multipartite entanglement test measuring how well a device builds genuine GHZ states, a transverse-field Ising model simulation for many-body dynamics, and a data re-uploading quantum neural network for classification. The entanglement and QNN tests include adjustable parameters, so a smaller or noisier device still runs a fair version.
Why this matters
QuSquare targets pre-fault-tolerant hardware, the machines available today, rather than hypothetical fault-tolerant systems. For readers running benchmarks themselves, see our randomized benchmarking guide and quantum volume tutorial. The suite is a research tool, not a product, and its value depends on adoption by hardware teams.