NVIDIA released the Ising model family, a set of open-source AI models targeting two core quantum computing bottlenecks. Ising Calibration is a 35B-parameter vision-language model trained on multi-modal qubit data that outperforms GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro on the new QCalEval benchmark by automating quantum processor tuning. Ising Decoding is a pair of 3D convolutional neural network models (0.9M and 1.8M parameters) for real-time quantum error correction, running 2.5× faster and 3× more accurately than pyMatching, the current open-source standard. Models are available on GitHub, Hugging Face, and build.nvidia.com. Adopters in the initial release include Fermilab, Harvard SEAS, IonQ, and the UK National Physical Laboratory.