Nvidia unveils Ising open source AI suite for quantum calibration
Summarize this article with:
Share Copy Link Share on X Share on Linkedin Share on Facebook Nvidia’s Ising open models target calibration and error correction needs. Credit: Mijansk786/Shutterstock.com. Nvidia has released Ising, a set of open source quantum AI models that aim to advance the development of quantum processors for practical use cases. Ising represents the first family of open source AI models targeting the quantum domain. The models are designed to address challenges such as processor calibration and quantum error correction that limit the scalability and reliability of current quantum computing systems. Nvidia positions these models as crucial for turning present-day quantum processors into platforms capable of running useful applications. The Ising models seek to accelerate calibration and error correction, both key factors in achieving scalable quantum computing. The company claims that the Ising suite delivers up to 2.5 times faster performance and three times higher accuracy for quantum decoding tasks compared to existing standards like pyMatching. Nvidia founder and CEO Jensen Huang “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.” Ising Calibration forms part of the offering as a vision-language model capable of interpreting and responding to quantum processor measurements. According to Nvidia, this enables the automation of calibration, potentially reducing the time required from days to hours. Ising Decoding includes two variants of a 3D convolutional neural network, optimised for either speed or accuracy, to perform the real-time decoding associated with quantum error correction. The models are already said to be in use at a range of organisations, including quantum computing companies, academic institutions, and national laboratories. The Ising suite allows researchers and enterprise users to run models locally on their systems, allowing full control over data and reducing dependence on external infrastructure. Ecosystem adoption of the calibration and decoding models spans institutions across North America, Europe, and Asia. Nvidia said that it provides additional resources such as workflow documentation and training data, along with NIM microservices, giving organisations the tools to adapt the Ising models for specific hardware and use cases. These models and data are accessible on GitHub, supporting continued open development. The Ising AI models have been named after the Ising mathematical model, historically used as a framework to simplify complex physical systems, reflecting their focus on high-performance and scalable solutions to quantum computing problems. Ising complements the Nvidia CUDA-Q software platform for hybrid quantum-classical computing. It enables direct integration with the Nvidia NVQLink QPU-GPU hardware interconnect for real-time control and quantum error correction requirements. This combination provides organisations with a comprehensive workflow from calibration to operational use in advanced computing systems, said Nvidia. The release marks Ising’s addition to Nvidia’s wider range of open model offerings, which also include Nemotron for agentic AI, Cosmos for physical AI, Alpamayo for autonomous vehicles, Isaac GR00T for robotics, and BioNeMo for biomedical research. All of these models and supporting frameworks are available through GitHub for the research and developer communities. In December 2025, the company expanded its portfolio with the introduction of the Nemotron 3 family. This group of open models and libraries, provided in Nano, Super, and Ultra sizes, uses a hybrid latent mixture-of-experts architecture designed to enable transparent, efficient, and specialised multi-agent deployments across various industries. Sign up for our regular news round-up! Give your business an edge with our leading Tech Monitor Sign up News Foresite, Tanium partner on managed autonomous endpoint operations News Pony.ai rolls out PonyWorld 2.0 to boost self-driving technology Comment Enterprises are all in on AI for security but budgets aren’t keeping pace Comment Adversaries have under-protected APIs in their sights