Research
NVIDIA Ising, first open-source AI models for quantum computing
NVIDIA Ising offers two groundbreaking AI models for quantum computing: Ising Calibration and Ising Decoding. We explain what they do and why they represent a revolution.
Quantum computing has fascinated researchers and engineers for decades, but it still faces colossal technical obstacles. To deliver on their promises, quantum computers must overcome two major challenges: keeping their processors perfectly calibrated and correcting errors inherent to qubits in real-time.
Until now, no truly adapted AI tool existed to address these. NVIDIA has just filled this void with NVIDIA Ising, a family of open-source AI models specifically designed to accelerate the path towards truly useful quantum computers.
Introducing NVIDIA Ising, the world’s first open AI models to accelerate the path to useful quantum computers.
— NVIDIA Newsroom (@nvidianewsroom) April 14, 2026
Researchers and enterprises can now use AI-powered workflows for scalable, high-performance quantum systems with quantum processor calibration capabilities and quantum… pic.twitter.com/jiwigqsOX
NVIDIA Ising: What Exactly Is It?
Is a family of open-source AI models that brings dedicated AI tools to the NVIDIA quantum platform, making them accessible to the entire quantum ecosystem. Specifically, it consists of two distinct models, each targeting a critical problem in quantum computing.
The first, Ising Calibration, is a Vision Language Model (VLM) with 35 billion parameters. Unique in its category, it is trained to interpret experimental data from a quantum processor (QPU) and deduce the calibration actions to be taken. It outperforms all other models on a suite of seven calibration performance tests and can interface with an agent to fully automate this process.
The second, Ising Decoding, is a duo of significantly lighter 3D convolutional neural network (3D CNN) models, with 0.9 or 1.8 million parameters. These models are optimized for pre-detection of quantum errors, leading to a 2.5x speed improvement and 3x precision improvement compared to the state of the art.
Why Are Calibration and Error Correction So Critical?
To understand the importance of NVIDIA Ising, two realities of quantum computing must be grasped. First, quantum processors require continuous adjustment to compensate for hardware imperfections. Current approaches are neither fast enough nor scalable enough; they still rely on human intervention or simplistic algorithms. With Ising Calibration, AI takes over to read the hardware status and automatically trigger the necessary corrections.
Second, for a quantum processor to function, qubit errors must be continuously corrected by quantum error correction codes. This involves processing terabytes of qubit measurement data, thousands of times per second, using demanding decoding algorithms. This is where Ising Decoding comes in, making this processing both faster and more reliable.
Without these two components, a quantum computer would remain unstable and unable to produce trustworthy results, which is currently one of the main obstacles to the democratization of the technology.
An Open Approach Integrated into the NVIDIA Ecosystem
What distinguishes NVIDIA Ising from other similar initiatives is its resolutely open-source nature and its deep integration into NVIDIA’s quantum platform. The models are published with permissive licenses, complete traceability of training data, and tools to retrain, fine-tune, or quantize them according to the specific needs of each team.
They are also provided pre-trained for common use cases, accompanied by practical guides to facilitate their adaptation. NVIDIA NIM microservices also enable instant deployment. NVIDIA Ising thus naturally integrates with other components of the company’s quantum platform, including CUDA-Q and NVQLink.
This openness is significant; it means that research labs, universities, and companies in the quantum sector can rely on these models without starting from scratch, thereby accelerating the entire field of innovation.
NVIDIA Ising and Ising Decoding: What to Remember?
With NVIDIA Ising, the semiconductor giant takes a decisive step by positioning AI as an indispensable lever for making quantum computing viable at scale. By opening these tools to the community, NVIDIA not only solves complex technical problems but also contributes to building the foundations of a new era of computing, where classical and quantum power will converge to solve problems currently out of reach.
[1] Cao, S., Pancotti, N., Lubowe, T., Svore, K., Kyoseva, E., Stanwyck, S., Costa, T., Zhang, Z., Mantilla Calderon, L., & Aspuru-Guzik, A. (2026, April 14). QCalEval: Benchmarking vision-language models for quantum calibration plot understanding. NVIDIA Research.
[2] Chamberland, C., Olle, J., Li, M., Thornton, S., & Baratta, I. (2026, April 14). Fast AI-based pre-decoders for surface codes. NVIDIA Research.
