Classiq and NVIDIA have demonstrated a working integration between the Classiq platform and NVIDIA's CUDA-Q framework, connecting high-level quantum circuit modeling with GPU-accelerated hybrid execution. In a published benchmark, a 31-qubit options pricing workflow that previously took approximately 67 minutes completed in roughly 2.5 minutes on a single NVIDIA A100 GPU.
The integration is available now through the Classiq development environment.
What the integration does
Classiq's platform operates at the model level. Developers describe quantum algorithms functionally rather than constructing circuits gate by gate. The platform synthesizes optimized circuits based on constraints like qubit count and target hardware compatibility.
CUDA-Q, NVIDIA's hybrid quantum-classical framework, handles execution. It allows quantum kernels to run alongside classical code while drawing on GPU acceleration, which happens to be especially useful for hybrid algorithms that invoke quantum circuits repeatedly inside classical optimization loops.
The integration connects these two stages directly. Developers can generate CUDA-Q kernels from Classiq programs and incorporate them into CUDA-Q hybrid workflows without manually reconstructing circuits at the lower level.
Why iteration speed matters for hybrid algorithms
Hybrid quantum algorithms such as VQE, QAOA, Iterative Quantum Amplitude Estimation, are structurally iterative. A classical optimizer drives repeated quantum circuit executions, adjusting parameters between runs. In practice, a single algorithm development session can involve hundreds or thousands of circuit evaluations.
When each evaluation is slow, the feedback loop stretches. Teams test fewer variants, explore parameter spaces less thoroughly, and identify bottlenecks later. The 27x reduction demonstrated in the IQAE benchmark is significant because experiments that required significant waiting can now run inside a normal development session.
The benchmark used an options pricing circuit synthesized in Classiq, executed via CUDA-Q on an A100. The problem domain, financial modeling with IQAE, is one where hybrid quantum algorithms are actively being studied, making it a reasonably representative test case rather than a synthetic one.
What developers can do with this today
The CUDA-Q integration is accessible through the Classiq platform. The workflow is:
Model your quantum algorithm in Classiq using its functional modeling tools
Export as a CUDA-Q kernel
Integrate the kernel into a CUDA-Q hybrid program alongside classical control logic or optimization routines
Execute in a CUDA-Q environment with GPU acceleration
Classiq's documentation includes a step-by-step walkthrough for the integration. Access to CUDA-Q environments requires NVIDIA GPU infrastructure, locally or via cloud.
What to watch
The benchmark covers simulation. Performance on real QPU hardware through CUDA-Q is a separate question the announcement does not address. Developers targeting actual quantum processors rather than GPU-simulated circuits should evaluate that pathway independently before drawing conclusions from the simulation figures.
The integration also currently flows one direction. Classiq models out to CUDA-Q kernels. Whether the reverse, CUDA-Q programs pulling Classiq synthesis inline, s on the roadmap is not stated.
For teams already using Classiq for algorithm design and evaluating hybrid execution environments, this is a concrete integration worth testing now. For those not yet in the Classiq ecosystem, the benchmark makes a case for the combination but adoption would involve evaluating both platforms independently.