Mark RadarMARK RADAR
About
EN
Sign in

NVIDIA Unveils Open-Source Ising Quantum AI Models; Jensen Huang Calls AI the Operating System for Quantum Machines

4 reports · First detected 2026-04-15 · Last active 2026-04-18

Qubits are highly susceptible to noise, requiring processors to be continuously calibrated and errors to be decoded in real time. Both tasks are major bottlenecks to scaling quantum computers and achieving fault-tolerant computing. NVIDIA is therefore integrating AI into quantum-GPU supercomputing platforms. CEO Jensen Huang said AI would become the control plane and operating system for quantum machines.

On World Quantum Day, April 14, 2026, NVIDIA unveiled its open-source Ising family of models, initially comprising Ising Calibration and Ising Decoding. Company data showed error-correction decoding speeds of up to 2.5 times those of conventional methods and accuracy of up to three times. Adopters include Academia Sinica, the Harvard John A. Paulson School of Engineering and Applied Sciences, and Fermi National Accelerator Laboratory.

All Coverage

4 original reports

The Backstory

The history behind this event
NVIDIA Open-Sources 31B Ising Model to Automate Quantum Calibration2026-07-28 · 1 reports · similarity 0.88

Quantum processing units require frequent calibration to keep fragile qubits operating consistently, a labor-intensive task that depends heavily on specialists interpreting experimental plots and adjusting control parameters. NVIDIA’s Ising Calibration model applies vision-language AI to those measurements, producing technical assessments, parameter extractions and experiment classifications. Automating that workflow could shorten maintenance cycles and help laboratories scale quantum systems toward fault-tolerant operation.

NVIDIA released the roughly 31-billion-parameter Ising Calibration 1.5 on July 23, 2026, with BF16 and new NVFP4 variants and support for multiple qubit technologies. The model analyzes calibration images and returns structured results through an OpenAI-compatible API. The NVFP4 profile can run on one supported GPU, reducing local deployment requirements for research laboratories. NVIDIA disclosed no investment amount or commercial transaction tied to the open release.

Mark Radar|MARK RADAR

If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →

All times are in Taipei time (GMT+8)