NVIDIA Launches Ising, World's First Open AI Models for Quantum Error Correction and Calibration

NVIDIA has unveiled Ising, a groundbreaking open-source family of AI models designed to solve two of quantum computing's most critical engineering challenges: processor calibration and error correction. Named after a landmark physics model, Ising provides high-performance, scalable AI tools that act as the 'operating system' for quantum machines. The Ising Decoding model, a 3D convolutional neural network, performs real-time quantum error correction up to 2.5x faster and 3x more accurately than the current industry standard, pyMatching. Meanwhile, the Ising Calibration model uses vision-language AI to automate the continuous tuning of quantum processors, slashing calibration time from days to mere hours.

This launch represents a strategic move to position AI as the essential control plane for the nascent quantum industry, which is projected to surpass $11 billion by 2030. NVIDIA is providing the models as NIM microservices alongside a cookbook of workflows, enabling researchers to fine-tune them for specific hardware while keeping proprietary data secure on their own systems. The Ising family integrates with NVIDIA's existing CUDA-Q software platform and NVQLink hardware interconnect, offering a full-stack solution for hybrid quantum-classical computing. Major adoption is already underway, with leading entities like Harvard's SEAS, Fermi National Accelerator Laboratory, Infleqtion, IQM Quantum Computers, and Sandia National Laboratories deploying the models to accelerate their path toward practical, large-scale quantum applications.