Public GitHub footprint of NVIDIA Corporation

Run agents like Hermes, LangChain Deep Agents, and OpenClaw more securely inside NVIDIA OpenShell with managed inference

Ongoing research training transformer models at scale

NVIDIA Linux open GPU kernel module source

TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT LLM also contains components to create Python and C++ runtimes that orchestrate the inference execution in a performant way.

Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, and security risks.

NVIDIA® TensorRT™ is an SDK for high-performance deep learning inference on NVIDIA GPUs. This repository contains the open source components of TensorRT.

NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more.

PersonaPlex code.

CUDA Templates and Python DSLs for High-Performance Linear Algebra

Samples for CUDA Developers which demonstrates features in CUDA Toolkit

A PyTorch Extension: Tools for easy mixed precision and distributed training in Pytorch

the LLM vulnerability scanner

OpenShell is the safe, private runtime for autonomous AI agents.

NVIDIA Isaac GR00T N1.7 - A Foundation Model for Generalist Robots.

A Python framework for GPU-accelerated simulation, robotics, and machine learning.

Transformer related optimization, including BERT, GPT

A GPU-accelerated library containing highly optimized building blocks and an execution engine for data processing to accelerate deep learning training and inference applications.

Tacotron 2 - PyTorch implementation with faster-than-realtime inference

Optimized primitives for collective multi-GPU communication

Build and run containers leveraging NVIDIA GPUs

NVIDIA device plugin for Kubernetes

A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on Hopper, Ada and Blackwell GPUs, to provide better performance with lower memory utilization in both training and inference.

CUDA Python: Performance meets Productivity

A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks like TensorRT-LLM, TensorRT, vLLM, etc. to optimize inference speed.

Pytorch implementation of FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks

Open-source deep-learning framework for building, training, and fine-tuning deep learning models using state-of-the-art Physics-ML methods

NeMo Retriever Library is a scalable, performance-oriented document content and metadata extraction microservice. NeMo Retriever Library uses specialized NVIDIA NIM microservices to find, contextualize, and extract text, tables, charts and images that you can use in downstream generative applications.

Minkowski Engine is an auto-diff neural network library for high-dimensional sparse tensors

NVIDIA GPU Operator creates, configures, and manages GPUs in Kubernetes

Agent Skills for NVIDIA products — install into Claude Code, Codex, and other coding agents to run Physical AI, robotics, simulation, CUDA, and RAG workflows end to end.

The NVIDIA NeMo Agent toolkit is an open-source library for efficiently connecting and optimizing teams of AI agents.

CUDA Library Samples

CUDA Core Compute Libraries

`std::execution`, the standard C++ framework for asynchronous and parallel programming.

cuTile is a programming model for writing parallel kernels for NVIDIA GPUs

TensorRT Extension for Stable Diffusion Web UI

AIStore: scalable storage for AI applications

NVIDIA curated collection of educational resources related to general purpose GPU programming.

NVIDIA GPU metrics exporter for Prometheus leveraging DCGM

NCCL Tests

This repo contains the source code for RULER: What’s the Real Context Size of Your Long-Context Language Models?

Tools for building GPU clusters

An efficient C++20 GPU numerical computing library with Python-like syntax

NVIDIA DLSS is a new and improved deep learning neural network that boosts frame rates and generates beautiful, sharp images for your games

A fast GPU memory copy library based on NVIDIA GPUDirect RDMA technology

Documentation of NVIDIA chip/hardware interfaces

LLM KV cache compression made easy

NVIDIA container runtime library

C++ and Python support for the CUDA Quantum programming model for heterogeneous quantum-classical workflows

A Python library that enables the use of Jetson's GPIOs

Open-source deep-learning framework for exploring, building and deploying AI weather/climate workflows.

RAFT contains fundamental widely-used algorithms and primitives for machine learning and information retrieval. The algorithms are CUDA-accelerated and form building blocks for more easily writing high performance applications.

Official Codebase for "DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos" (ICML 2026)

NVIDIA cuDF for Apache Spark plugin - accelerate Apache Spark with GPUs

GPU accelerated decision optimization

NVIDIA Federated Learning Application Runtime Environment

CUDA Kernel Benchmarking Library

cuDNN Frontend is NVIDIA's modern, open-source entry point to the cuDNN library and a growing collection of high-performance open-source kernels.

cuVS - a library for vector search and clustering on the GPU

high-performance inference and serving library for interactive autoregressive video and world models

JAX-Toolbox

Our inference and training framework to run on the Cosmos Models

repo collection for NVIDIA Audio2Face-3D models and tools

Megatron's multi-modal data loader

Tooling for optimized, validated, and reproducible GPU-accelerated AI runtime in Kubernetes

NVSentinel is a cross-platform fault remediation service designed to rapidly remediate runtime node-level issues in GPU-accelerated computing environments

The unified framework for sim & real robot teleoperation

NVIDIA Resiliency Extension is a python package for framework developers and users to implement fault-tolerant features. It improves the effective training time by minimizing the downtime due to failures and interruptions.

NVIDIA OptiX based implementation of ANARI

NVIDIA Kaggle Plugin gives agents end-to-end Kaggle competition workflows through a single skill, nvidia-kaggle-skill. It can gather competition context, study public writeups and notebooks, reproduce kernels locally, submit to competitions, and manage Ka

Help shape the future of Project G-Assist

NVIDIA Infra Controller - Hardware Lifecycle Management and multitenant networking

Cosmos Curator is a powerful video curation system that processes, analyzes, and organizes video content using advanced AI models and distributed computing.

ALCHEMI Toolkit-Ops is a collection of optimized batch kernels to accelerate computational chemistry and material science workflows.

The developer-first platform for scaling complex Physical AI workloads across heterogeneous compute—unifying training GPUs, simulation clusters, and edge devices in a simple YAML

An Online Deep Learning Interface for HPC programs on NVIDIA GPUs

Platform for deploying and routing GPU-accelerated inference, streaming, and batch workloads at scale.

A repo for all spark examples using Rapids Accelerator including ETL, ML/DL, etc.

NVIDIA vGPU Device Manager manages NVIDIA vGPU devices on top of Kubernetes

A toolkit for discovering cluster network topology.

Audio-to-Audio Schrodinger Bridges is a diffusion-based audio restoration model for bandwidth extension and inpainting.

Set of utilities supporting workflows common in GPU raytracing applications

Kubernetes Operator, Helm Charts, Ansible Playbooks, and utility scripts for large-scale AIStore deployments on Kubernetes.

InstantNuRec: Feed-Forward 3D Gaussian Reconstruction from Driving Logs

A set of training recipes for AI Quantum Error Correction Decoders

Ubuntu kernels which are optimized for NVIDIA server systems

Accelerated libraries for quantum-classical computing built on CUDA-Q.

DOCA Platform manages provisioning and service orchestration for Bluefield DPUs

Multi-language agent runtime and library for execution scope management, lifecycle events, and middleware on tool and LLM calls.

XR AI

RAPIDS Accelerator JNI For Apache Spark

NVIDIA Halos Outside-In Safety Blueprint extends robot perception beyond on-board sensors by using external infrastructure cameras and AI agents to dynamically control robot behavior and perform at maximum efficiency.

K8s-test-infra

NVIDIA Inference Benchmarks provide recipes in ready-to-use templates for evaluating platform speed. Validate your platform across specific AI use cases across hardware and software combinations.

NVIDIA's Redfish next generation redfish crate

Linux kernel source tree

NVIDIA Rack Management Service Rust language client crate

A diagnostic framework that decomposes 3D object counting into nine hierarchical spatial perception and cognition sub-tasks. Analysis code for the paper Spatial-IQ: Deconstructing Spatial Intelligence via Hierarchical Capability Tests.

🧪 OpenShell's Research Journal

MCTP tool running on Host to communicate with devices via USB and other transport layers.