# NVIDIA AI Foundation Models Reasoning and instruct models optimized for enterprise agentic AI. [Explore Foundation Models](https://build.nvidia.com/explore/discover) ## What Are NVIDIA AI Foundation Models? NVIDIA AI foundation models are community- and NVIDIA-built models that are optimized to deliver the best performance on NVIDIA-accelerated infrastructure. Enterprises can customize and deploy these models with NVIDIA NIMTM microservices and streamline the transition to production AI. Explore the [NVIDIA API catalog](http://build.nvidia.com) and experience the multimodal models directly from a browser, connect to NVIDIA-hosted endpoints and start POC for free, or download and run on your compute platform. ## Supercharge Agents With Expert Reasoning Power your enterprise agents with open [NVIDIA Llama Nemotron with reasoning models](https://nvidianews.nvidia.com/news/nvidia-launches-family-of-open-reasoning-ai-models-for-developers-and-enterprises-to-build-agentic-ai-platforms?nvid=nv-int-bnr-616981) that deliver leading accuracy  and efficiency for a wide range of reasoning and non-reasoning agentic tasks. Available as NVIDIA NIM microservices, the NVIDIA-developed and 3rd party models are optimized for performance and can be rapidly deployed on any NVIDIA-accelerated infrastructure. [Explore Reasoning Models](https://www.nvidia.com/en-us/ai-data-science/foundation-models/llama-nemotron.md) # NVIDIA NeMo A comprehensive toolkit to build, monitor, and optimize AI agents across their lifecycle at enterprise scale. [Get Started](https://www.nvidia.com/en-us/ai-data-science/products/nemo/get-started.md) [Documentation](https://docs.nvidia.com/nemo/) ## The World Foundation Model Platform to Accelerate Physical AI The development of physical-AI-embodied systems such as robots and autonomous vehicles is accelerated with the new NVIDIA CosmosTM platform. [Read the Press Release](https://nvidianews.nvidia.com/news/nvidia-launches-cosmos-world-foundation-model-platform-to-accelerate-physical-ai-development) ## Build Custom AI Models for Enterprise AI Agents The [NVIDIA AI Foundry](https://www.nvidia.com/en-us/ai/foundry.md) —a collection of NVIDIA AI foundation models, [NVIDIA NeMoTM](https://www.nvidia.com/en-us/ai-data-science/generative-ai/nemo-framework.md) framework and microservices, [NVIDIA NIM microservices](https://www.nvidia.com/en-us/ai-data-science/products/nim-microservices.md), and [NVIDIA DGXTM Cloud](https://www.nvidia.com/en-us/data-center/dgx-cloud.md) —gives enterprises an end-to-end solution for creating custom generative AI models. ### Start with State-of-the-Art Generative AI Models Try leading multimodal models, including Llama, Llama Nemotron with reasoning, Cosmos Nemotron, and Phi, optimized for the highest performance and efficiency. [Experience NVIDIA AI Foundation Models](https://build.nvidia.com/explore/discover) ### Customize the Foundation Models Tune and test the models with proprietary data using NVIDIA NeMo. [Customize With NVIDIA NeMo](https://www.nvidia.com/en-us/ai-data-science/generative-ai/nemo-framework.md) ### Build Models Faster in the Cloud Customize models on DGX Cloud, a serverless AI-training-as-a-service platform for enterprise developers. [Train on NVIDIA DGX Cloud](https://www.nvidia.com/en-us/data-center/dgx-cloud.md) ### Run Models in Production Deploy custom and NVIDIA AI foundation models anywhere with enterprise-grade NVIDIA NIM. [Scale With NVIDIA NIM](https://build.nvidia.com/explore/discover) ## Benefits of NVIDIA AI Foundation Models and Endpoints ### Performance Optimized Lower your TCO and increase energy efficiency by running inference up to 4x faster. ### Enterprise-Grade Use lean, high-performing LLMs built from responsibly sourced datasets. ### Try Models on the Fly Experience a model’s peak performance directly from a browser with a GUI or API. ### Ready-to-Integrate APIs Connect your applications to API endpoints and test their real-world performance running on a fully-accelerated stack. ### Deploy Your Models Anywhere Run the model anywhere, from cloud to data center to workstations, with NVIDIA AI Enterprise. ## Experience-Optimized Generative AI Models NVIDIA AI foundation models include leading community- and NVIDIA-built models to support various use cases, including content generation, image creation, drug discovery, and IT service automation. ### Llama Nemotron Highest reasoning accuracy and efficiency models to advance autonomy of complex agentic AI systems. [View Llama Nemotron Models](https://www.nvidia.com/en-us/ai-data-science/foundation-models/llama-nemotron.md) ### Cosmos [NVIDIA CosmosTM](https://www.nvidia.com/en-us/ai/cosmos.md), a platform of state-of-the-art generative [world foundation models](https://www.nvidia.com/en-us/glossary/world-models.md), advanced tokenizers, guardrails, and an accelerated data processing and curation pipeline built to accelerate the development of [physical AI](https://www.nvidia.com/en-us/glossary/physical-ai.md) systems such as autonomous vehicles (AVs) and robots. [Try Cosmos](https://developer.nvidia.com/cosmos) ### Mistral Large Mistral Large excels in complex multilingual reasoning tasks, including text understanding, and code generation. [Try Mistral Large](https://build.nvidia.com/mistralai/mistral-large) [View All Models](https://build.nvidia.com/explore/discover) ## Power Your Enterprise Applications With Retrieval-Augmented Generation (RAG) Build AI chatbots that connect with your custom LLMs and knowledge bases to accurately and naturally answer domain-specific questions in real time. [Explore the RAG AI Workflow](https://www.nvidia.com/en-us/ai-data-science/ai-workflows/generative-ai-chatbots.md) ## Success Stories Generative AI is impacting every industry today—from IT services and telecommunications to finance and retail.  Putting generative AI into practice requires enterprises to have access to an AI foundry to build custom models using proprietary data and deploy them at scale. See how the world’s leading organizations are serving their customers with NVIDIA AI. ### ServiceNow ServiceNow is bringing intelligent workflow automation to their Now Platform with custom LLMs using NVIDIA AI foundation models and NVIDIA NeMo on NVIDIA DGX. [Learn More](https://nvidianews.nvidia.com/news/servicenow-and-nvidia-announce-partnership-to-build-generative-ai-across-enterprise-it/) ### Amdocs Amdocs is building custom LLMs for the $1.7 trillion global telecommunications industry using the NVIDIA AI foundry service on Microsoft Azure. [Learn More](https://blogs.nvidia.com/blog/2023/11/15/amdocs-telco-industry/) Previous Next 1. cont-1 2. cont-2 ### Ecosystem Partners ### Let’s Get Started Try the latest, fully optimized NVIDIA AI foundation models today from the NGC catalog, Azure ML model catalog, or Hugging Face. [Experience the Models](https://build.nvidia.com/explore/discover) Notify me as new models are optimized and added to NVIDIA’s collection of AI foundation models. [Notify Me](https://www.nvidia.com/en-us/ai-data-science/generative-ai/ai-foundations-model-notify-me.md) Explore additional generative AI resources and tools. [Learn More](https://resources.nvidia.com/en-us-generative-ai-getting-started)