NVIDIA has launched Ising, the world’s first open-source quantum AI model family, targeting calibration and error correction bottlenecks that stand between today’s qubits and useful quantum systems.
NVIDIA has unveiled Ising, the world’s first family of open-source quantum AI models, positioning open-source AI as the control layer for fault-tolerant quantum computing infrastructure.
The launch directly addresses the two biggest engineering roadblocks in scaling quantum systems—processor calibration and quantum error correction—with performance claims of up to 2.5x faster decoding and 3x higher accuracy than pyMatching, the current open-source benchmark.
The Ising family includes Ising Calibration, a vision-language model that automates processor tuning, and Ising Decoding, two 3D convolutional neural network variants optimised for either speed or accuracy in real-time quantum error correction. NVIDIA said the calibration workflow can now shrink from days to hours through AI agents performing continuous automated tuning.
“AI is essential to making quantum computing practical. With Ising, AI becomes the control plane — the operating system of quantum machines — transforming fragile qubits to scalable and reliable quantum-GPU systems,” said Jensen Huang, founder and CEO, NVIDIA.
The open-source release arrives with a quantum workflow cookbook, training datasets, NVIDIA NIM microservices, local deployment support, and hardware-specific fine-tuning tools, enabling researchers to adapt models while preserving proprietary data.
Adoption at launch already spans Academia Sinica, Fermi National Accelerator Laboratory, Harvard, IonQ, IQM Quantum Computers, Sandia National Laboratories, UC San Diego, and Yonsei University, among others.
Integrated with CUDA-Q and NVQLink, and distributed through GitHub, Hugging Face, and build.nvidia.com, Ising marks a major open-source infrastructure play as the quantum computing market races toward $11 billion by 2030.
