NVIDIA Launches Ising Open Models for Quantum Computing

NVIDIA has announced a new family of open models called NVIDIA Ising, designed to address quantum processor calibration and quantum error correction. These are two of the main engineering challenges limiting the scalability of current quantum systems, where noise and instability in qubits reduce the reliability of computations. The Ising models are intended to automate parts of this process using machine learning, enabling faster calibration cycles and more efficient decoding of quantum errors during execution.

The Ising family includes two main components. The calibration model is a vision-language system that interprets measurement data from quantum hardware and adjusts parameters in near real time, reducing manual intervention and shortening calibration cycles. The decoding models are based on 3D convolutional neural networks that process error syndromes for quantum error correction, with variants optimized for either latency or accuracy. According to NVIDIA, these models can outperform existing approaches such as pyMatching in both speed and accuracy, enabling more practical real-time error correction workflows.

The models are released as open source and can be deployed locally or adapted to specific quantum hardware setups. NVIDIA is also providing supporting datasets, workflow examples, and NIM microservices to help developers integrate and fine-tune the models. The system integrates with CUDA-Q for hybrid quantum-classical programming and NVQLink for…