NVIDIA Introduces Ising Open Models for Quantum Computing

NVIDIA has launched Ising, a family of open-source AI models designed to accelerate progress toward practical quantum computing, announced in a press release. The Ising models aim to enhance quantum processor calibration and error correction, two key challenges in building scalable hybrid quantum-classical systems.

The Ising suite includes two core tools: Ising Calibration, a vision-language model that automates continuous quantum processor calibration, and Ising Decoding, a 3D convolutional neural network optimized for quantum error correction. NVIDIA reports that Ising Decoding performs up to 2.5 times faster and with three times greater accuracy than pyMatching, the current open-source standard.

Early adopters include Atom Computing, Academia Sinica, Harvard University, IQM Quantum Computers, and the U.K. National Physical Laboratory. The models integrate with NVIDIA’s CUDA-Q software platform and NVQLink hardware interconnect, allowing real-time control and error correction between GPUs and quantum processing units.

The Ising models are available through GitHub, Hugging Face, and build.nvidia.com, joining NVIDIA’s portfolio of open AI tools such as Nemotron for agentic systems, Cosmos for physical AI, and BioNeMo for biomedical research.