NVIDIA on September 14, 2026, expanded its open-source CUDA-Q platform with CUDA-Q Logical, an orchestration layer for designing and testing fault-tolerant quantum computing applications. The company said the layer is now available through GitHub, while a reference implementation of Sandia National Laboratories’ new QUOPS benchmark is available in CUDA-Q.
CUDA-Q Logical is intended to let researchers coordinate algorithms, error-correction codes, hardware architectures and other quantum processing unit components, then switch among options to compare configurations. NVIDIA said this approach addresses a codesign problem in which changing one component can substantially alter the resources required for an application. Fault-tolerant processors use logical qubits to overcome errors in physical qubits and support larger computations, including potential work in drug discovery, financial modeling and materials development.
Fermi National Accelerator Laboratory used CUDA-Q Logical to validate previous results and evaluate physical-qubit counts, runtimes and other resource requirements across error-correction methods and hardware. According to NVIDIA, Fermilab converted the work into a repeatable, verifiable computational workflow and reduced a process that typically took about five months to three weeks, described as a sevenfold speedup. CUDA-Q Logical is also being used by Infleqtion, IQM Quantum Computers, QCDesign, Quantum Motion and Sandia National Laboratories.
In another project, Iceberg Quantum modeled its fault-tolerant architecture for Diraq’s qubits. The modeling indicated that 1,000 logical qubits could be created with 150,000 physical qubits, roughly 10 times fewer than Diraq’s previous estimates, according to NVIDIA. The company said CUDA-Q Logical enabled rapid assessment of possible implementations of Iceberg’s architecture.
QUOPS, developed independently by Sandia, is an open, hardware-agnostic benchmark designed to measure progress toward utility-scale, fault-tolerant quantum applications. It shifts attention beyond conventional measures such as physical-qubit counts, fidelity and coherence by tracking and forecasting quantum-computer capabilities for practical workloads. Sandia reported initial QUOPS results for quantum processors from Google, IBM and Quantinuum in a preprint posted ahead of IEEE Quantum Week.
NVIDIA also detailed broader adoption of its quantum-computing technologies. Diraq used NVIDIA Ising models to calibrate a silicon-based qubit processor, while Anyon Computing, Quandela and Quantum Machines used the NVQLink architecture in quantum processor and GPU-supercomputing projects. BlueQubit launched a grant program providing researchers access to NVIDIA accelerated computing through CUDA-Q, and Qedma Quantum Computing and QCentroid integrated their technologies with the platform for error correction, mitigation and quantum-application deployment.