NVIDIA Technical Blog

The latest release of the NVIDIA Collective Communications Library (NCCL) introduces a groundbreaking fusion of communication and computation for higher...

On November 18, learn how to fine-tune the NVIDIA Cosmos Reason VLM with your own data to create visual AI agents.

The NVIDIA Collective Communications Library (NCCL) provides communication APIs for low-latency and high-bandwidth collectives, enabling AI workloads to scale...

One of the many strengths of the PyData ecosystem is interoperability, which enables seamlessly moving data between libraries that specialize in exploratory...

As AI weather and climate prediction models rapidly gain adoption, the NVIDIA Earth-2 platform provides libraries and tools for accelerating solutions using a...

Large language models can solve challenging math problems. However, making them work efficiently at scale requires more than a strong checkpoint. You need the...

The NVIDIA GB200 NVL72 pushes AI infrastructure to new limits, enabling breakthroughs in training large-language models and running scalable, low-latency...

Over the past few years, AI inference has evolved from single-model, single-pod deployments into complex, multicomponent systems. A model deployment may now...

Modern AI workloads have moved well beyond single-GPU inference serving. Model parallelism, which efficiently splits computation across many GPUs, is now the...

SemiAnalysis recently launched InferenceMAX v1, a new open source initiative that provides a comprehensive methodology to evaluate inference hardware...

The exponential growth in large language model complexity has created challenges, such as models too large for single GPUs, workloads that demand high...

As AI models grow larger and more sophisticated, inference, the process by which a model generates responses, is becoming a major challenge. Large language...

Generating text with large language models (LLMs) often involves running into a fundamental bottleneck. GPUs offer massive compute, yet much of that power sits...

Deploying large language models (LLMs) poses a challenge in optimizing inference efficiency. In particular, cold start delays—where models take significant...

The race to understand protein structures has never been more critical. From accelerating drug discovery to preparing for future pandemics, the ability to...

AI models, inference engine backends, and distributed inference frameworks continue to evolve in architecture, complexity, and scale. With the rapid pace of...

Data scientists spend a lot of time cleaning and preparing large, unstructured datasets before analysis can begin, often requiring strong programming and...

What if you could talk to your computer and have it perform tasks through the Bash terminal, without you writing a single command? With NVIDIA Nemotron Nano v2,...

Modern organizations generate a massive volume of operational data through ticketing systems, incident reports, service requests, support escalations, and more....

Logs are the lifeblood of modern systems. But as applications scale, logs often grow into endless walls of text—noisy, repetitive, and overwhelming. Hunting...

Unlike traditional LLM-based systems that are limited by their training data, retrieval-augmented generation (RAG) improves text generation by incorporating...

Unlike traditional systems that follow predefined paths, AI agents are autonomous systems that use large language models (LLMs) to make decisions, adapt to...

AI agents now solve multi-step problems, write production-level code, and act as general assistants across multiple domains. But to reach their full potential,...

Have you ever wanted to build your own reasoning models such as the NVIDIA Nemotron, but thought it was too complicated or required massive resources? Think...

Can AI coding assistants write efficient CUDA code? To help measure and improve their capabilities, we created ComputeEval, a robust, open source benchmark for...

As companies collect more unstructured data and increasingly use large language models (LLMs), they need faster and more scalable systems. Advanced tools for...

Training massive mixture-of-experts (MoE) models has long been the domain of a few advanced users with deep infrastructure and distributed-systems expertise....

Organizations are increasingly seeking ways to extract insights from video, audio, and other complex data sources. Retrieval-augmented generation (RAG) enables...

Medical AI has reached an inflection point. While vision-language models (VLMs) have shown promise in medical imaging, they have lacked the systematic,...

AI-driven applications are evolving from passive tools to agentic systems that generate code, make decisions, and take autonomous actions. This shift introduces...

Modern AI workloads, ranging from large-scale training to real-time inference, demand dynamic access to powerful GPUs. However, Kubernetes environments have...

Autonomous vehicle (AV) stacks are evolving from a hierarchy of discrete building blocks to end-to-end architectures built on foundation models. This transition...

Traditional task and motion planning (TAMP) systems for robot manipulation use cases operate on static models that often fail in new environments. Integrating...

As robots take on increasingly dynamic mobility tasks, developers need physics-accurate simulations that scale efficiently across environments and workloads....

Building realistic 3D environments for robotics simulation can be a labor-intensive process. Now, with NVIDIA Omniverse NuRec, you can complete the entire...

A defining strength of the NVIDIA software ecosystem is its commitment to continuous optimization. In August, NVIDIA Jetson AGX Thor launched, with up to a 5x...

Training robot policies from real-world demonstrations is costly, slow, and prone to overfitting, limiting generalization across tasks and environments. A...

Physics plays a crucial role in robotic simulation, providing the foundation for accurate virtual representations of robot behavior and interactions within...

The increasing demand for robotics is driving the need for physics-accurate simulation at an unprecedented scale. Universal Scene Description (OpenUSD) is key...

Modern robotics requires more than what classical analytic dynamics provides because of simplified contacts, omitted kinematic loops, and non-differentiable...

Training models with billions or trillions of parameters demands advanced parallel computing. Researchers must decide how to combine parallelism strategies,...

​​For decades, one of biology’s deepest mysteries was how a string of amino acids folds itself into the intricate architecture of life. Researchers built...

Open research is critical for driving innovation, and many breakthroughs in AI and science are achieved through open collaboration. In the field of digital...

How does the NFL schedule all its regular-season games while avoiding stadium conflicts with Beyoncé concerts? How can doctors use a single donated...

Today’s demanding AI developer workloads often need more memory than desktop systems provide or require access to software that laptops or PCs lack. This...

Build the next generation of intelligent, autonomous applications. This isn't just a hackathon—it's your chance to unleash the power of agentic AI and show...

Built for data scientists and bioinformaticians, NVIDIA Parabricks is a scalable genomics software suite for secondary analysis. Providing GPU-accelerated...

Predicting where proteins are located inside a cell is critical in biology and drug discovery. This process is known as subcellular localization. The location...

Join the Developer Kernel Hackathon, a four-part performance challenge hosted by NVIDIA in collaboration with GPU MODE and support from Dell and Sesterce. Push...

NVIDIA CUDA-X math libraries provide the fundamental numerical building blocks that enable developers to deploy accelerated applications across multiple...

Download NERSC’s QGEAR project to easily import Qiskit circuits into GPU-accelerated CUDA-Q kernels.

Molecular dynamics (MD) simulations are a powerful tool in computational chemistry and materials science, and they’re essential for studying chemical...

While today's robots excel in controlled settings, they still struggle with the unpredictability, dexterity, and nuanced interactions required for real-world...

Scientists from NVIDIA, in collaboration with Lawrence Berkeley National Laboratory (Berkeley Lab), released a machine learning tool called Huge Ensembles...

The new release introduces CUDA 13.0 support and new functions for ahead-of-time compilation module.

Computer-aided engineering (CAE) forms the backbone for modern product development across industries, from designing safer aircraft to optimizing renewable...

Agentic AI is an ecosystem where specialized language and vision models work together. They handle planning, reasoning, retrieval, and safety guardrailing....

Generative AI is opening new possibilities for analyzing existing video streams. Video analytics are evolving from counting objects to turning raw video content...

Building a robust visual inspection pipeline for defect detection and quality control is not easy. Manufacturers and developers often face challenges such as...

Powerful hardware-accelerated video processing in Python just got easier. PyNvVideoCodec is an NVIDIA Python-based library for GPU-accelerated video encoding,...

The constantly increasing compute throughput of NVIDIA GPUs presents a new opportunity for optimizing vision AI workloads: keeping the hardware fed with data....

Robotics is undergoing a revolution, moving beyond the era of specialist machines to generalist robotics. This shift moves away from single-purpose,...

First unveiled at NVIDIA GTC 2025, NVIDIA Cosmos Reason is an open and fully customizable reasoning vision language model (VLM) for physical AI and robotics....

Human action recognition is a capability in AI systems designed for safety-critical applications, such as surveillance, eldercare, and industrial monitoring....

To help create real-time, dynamic NPC game characters, NVIDIA ACE now supports the open source Qwen3-8B small language model (SLM) for on-device...

DOOM: The Dark Ages pushes real-time graphics to new limits by integrating RTX neural rendering and path tracing, setting a new standard for how modern games...

By leveraging large language and speech models, generative AI is creating intelligent 3D avatars that can engage users in natural conversation, from video games...

Running 3D applications that take advantage of advanced rendering and simulation technologies often requires users to navigate complex installs and have access...

Today at Gamescom 2025, NVIDIA unveiled updates to NVIDIA RTX neural rendering and NVIDIA ACE generative AI technologies that enable developers to deliver...

Transferring 3D data between applications has long been a challenge, especially with proprietary formats such as native computer-aided design (CAD) files. CAD...

The proliferation of generative AI video models, along with the new workflows these models have introduced, has significantly accelerated production efficiency...

As part of continued efforts to ensure NVIDIA Omniverse is a developer-first platform, NVIDIA will be deprecating the Omniverse Launcher on Oct. 1. Doing so...

Wireless communication research is rich with brilliant ideas and computational power. Yet, there's a fundamental disconnect between what researchers can...

The telecommunications industry is innovating rapidly toward 6G for both AI-native Radio Access Networks (AI-RAN) and AI-Core. The distributed User Plane...

The world of embedded and edge computing is about to get faster, more efficient, and more versatile with the upcoming CUDA 13.0 release for Jetson Thor SoC...

Telesurgery is no longer a futuristic idea—it’s quickly becoming essential to how care is delivered. With a global shortage of surgeons projected to reach...

By 2030, the World Health Organization projects a global shortage of over 15 million healthcare workers, including surgeons, radiologists, and nurses. In the...

Researchers have developed an AI-powered tool that can analyze nurses’ shift notes to identify—far earlier than traditional methods—when an admitted...

As industrial automation accelerates, factories are increasingly relying on advanced robotics to boost productivity and operational resilience. The successful...

NVIDIA TensorRT for RTX is now available for download as an SDK that can be integrated into C++ and Python applications for both Windows and Linux. At...

Fine-tuning and reinforcement learning (RL) for large language models (LLMs) require advanced expertise and complex workflows, making them out of reach for...

If you're an application developer or a cluster administrator, you’ve likely seen how non-uniform memory access (NUMA) can impact system performance. When an...

For decades, traditional data centers have been vast halls of servers with power and cooling as secondary considerations. The rise of generative AI has changed...

NVIDIA HGX B200 is revolutionizing accelerated computing by unlocking unprecedented performance and energy efficiency. This post shows how HGX B200 is...

Ultra-low latency and reliable packet delivery are critical requirements for modern applications in sectors such as the financial services industry (FSI), cloud...

Building and deploying applications can be challenging for developers, requiring them to navigate the complex relationship between hardware and software...

AI scaling is incredibly complex, and new techniques in training and inference are continually demanding more out of the data center. While data center...

As large language models (LLMs) grow larger, they get smarter, with open models from leading developers now featuring hundreds of billions of parameters. At the...

NVIDIA KAI Scheduler is now natively integrated with KubeRay, bringing the same scheduling engine that powers high‐demand and high-scale environments in...

In AI infrastructure, data fuels the compute engine. With evolving agentic AI systems, where multiple models and services interact, fetch external context, and...

NVIDIA is reshaping the landscape of data-center connectivity by seamlessly integrating optical and electrical components. But it’s not doing it alone....

As the latest member of the NVIDIA Blackwell architecture family, the NVIDIA Blackwell Ultra GPU builds on core innovations to accelerate training and AI...

The exponential growth in AI model complexity has driven parameter counts from millions to trillions, requiring unprecedented computational resources that...

As artificial intelligence redefines the computing landscape, the network has become the critical backbone shaping the data center of the future. Large language...

Continuous integration and continuous delivery/deployment (CI/CD) is a set of modern software development practices used for delivering code changes more...

The NVIDIA Collective Communications Library (NCCL) is essential for fast GPU-to-GPU communication in AI workloads, using various optimizations and tuning to...