NVIDIA Ising Introduces AI-Powered Workflows to Build Fault-Tolerant Quantum Systems

NVIDIA Ising is the world’s first family of open AI models for building quantum processors, launching with two model domains: Ising Calibration and Ising Decoding. Both target the fundamental challenge in quantum computing—qubits are inherently noisy. The best quantum processors make an error roughly once in every thousand operations. To become useful accelerators for scientific and enterprise problems, error rates must drop to one in a trillion or better. AI is the most promising path to closing that gap at scale. Calibration is the process of understanding the noise in each quantum processor and tuning it to achieve the best possible performance. Calibration minimizes error, but because of noise in quantum systems, errors must be corrected in real time by a classical computer, faster than they accumulate. This process is called quantum error correction decoding. Both calibration and decoding are computationally intensive and need improved methods to drive progress. Ising delivers advanced performance on calibration and error correction decoding, using techniques for scaling to millions of qubits. NVIDIA Ising provides open base models, a training framework, and workflows for fine-tuning, quantization, and deployment. The pre-trained models deliver top performance out of the box, and because everything is open, users can also specialize for their own hardware and noise characteristics while keeping proprietary QPU data on-site. In this post, we dive into how NVIDIA Ising delivers starting points for users to select base models, train their own, fine-tune, quantize, and deploy optimized inference workflows wherever needed, improving QPU performance and providing a path to scale to Quantum-GPU Supercomputers capable of solving useful problems. The NVIDIA Ising family launches with two breakthrough models: - NVIDIA Ising Calibration: A vision-language model (VLM) model for automating QPU calibration tasks. - NVIDIA Ising Decoding: Consists of two 3D CNN models for demanding decoding needed during quantum error correction. NVIDIA Ising Calibration NVIDIA Ising Calibration is a VLM capable of understanding quantum computing scientific experiment output and how it compares to expected trends. This VLM can be used in an agentic workflow that responds to measurement results and actively calibrates a quantum processor until its operation falls within desired specifications. The Ising-Calibration-1 model was trained on data generated from information provided by partners spanning multiple qubit modalities, including superconducting qubits, quantum dots, ions, neutral atoms, electrons on Helium, and others specializing in calibration and control. In the absence of a standard benchmark for evaluating quantum calibration models, NVIDIA collaborated with quantum partners to develop QCalEval, the world’s first benchmark for agentic quantum computer calibration, containing real quantum computer outputs. This benchmark is a six-part semantic scoring test that assesses any model’s effectiveness at relevant calibration tasks. QCalEval measures a model’s ability to interpret experimental results, classify outcomes, evaluate their significance, assess fit quality and key features, and generate actionable next-step recommendations. Learn more about the QCalEval benchmark, along with model architecture and evaluation results Ising-Calibration-1 repeatedly outperforms state-of-the-art open and closed models of a range of parameters. As shown in Figure 1, Ising Calibration 1 scores…