Accelerate domain adaptation of LLMs.
DeepSeek-V3.1: hybrid inference LLM with Think/Non-Think modes, stronger agents, 128K context, strict function calling.
Multilingual, cross-lingual embedding model for long-document QA retrieval, supporting 26 languages.
Japanese-specialized large-language-model for enterprises to read and understand complex business documents.
Qwen3-Next Instruct blends hybrid attention, sparse MoE, and stability boosts for ultra-long context AI.
State-of-the-art model for Polish language processing tasks such as text generation, Q&A, and chatbots.
80B parameter AI model with hybrid reasoning, MoE architecture, support for 119 languages.
Record-setting accuracy and performance for Mandarin English transcriptions.
Accurate and optimized Vietnamese-English transcriptions with punctuation and word timestamps.
Transform your scene idea into ready-to-use 3D assets using Llama 3.1 8B, NV SANA, and Microsoft TRELLIS
Elevate Shopping Experiences Online and In Stores.
DeepSeek V3.1 Instruct is a hybrid AI model with fast reasoning, 128K context, and strong tool use.
High‑efficiency LLM with hybrid Transformer‑Mamba design, excelling in reasoning and agentic tasks.
Reasoning vision language model (VLM) for physical AI and robotics.
Powerful OCR model for fast, accurate real-world image text extraction, layout, and structure analysis.
Smaller Mixture of Experts (MoE) text-only LLM for efficient AI reasoning and math
Accurate and optimized English transcriptions with punctuation and word timestamps
Sensor-captured radio enables real-time awareness, AI-driven analytics for actionable, searchable insights.
High efficiency model with leading accuracy for reasoning, tool calling, chat, and instruction following.
Multilingual 7B LLM, instruction-tuned on all 24 EU languages for stable, culturally aligned output.
Multilingual, cross-lingual embedding model for long-document QA retrieval, supporting 26 languages.
Powerful OCR model for fast, accurate real-world image text extraction, layout, and structure analysis.
Expressive and engaging text-to-speech, generated from a short audio sample.
Translation model in 12 languages with few-shots example prompts capability.
Multi-modal model to classify safety for input prompts as well output responses.
Enable smooth global interactions in 36 languages.
An edge computing AI model which accepts text, audio and image input, ideal for resource-constrained environments
An edge computing AI model which accepts text, audio and image input, ideal for resource-constrained environments
GPU-accelerated model optimized for providing a probability score that a given passage contains the information to answer a question.
Generates physics-aware video world states for physical AI development using text prompts and multiple spatial control inputs derived from real-world data or simulation.
Removes unwanted noises from audio improving speech intelligibility.
Multimodal question-answer retrieval representing user queries as text and documents as images.
Use the multi-LLM compatible NIM container to deploy a broad range of LLMs from Hugging Face.
Build advanced AI agents within the biomedical domain using the AI-Q Blueprint and the BioNeMo Virtual Screening Blueprint
Build a data flywheel, with NVIDIA NeMo microservices, that continuously optimizes AI agents for latency and cost — while maintaining accuracy targets.
Streamline evaluation, monitoring, and optimization of AI data flywheel with Weights & Biases.
Orchestrate AI agents for data flywheel with MLRun and NVIDIA NeMo microservices.
Improve safety, security, and privacy of AI systems at build, deploy and run stages.
Automate and optimize the configuration of radio access network (RAN) parameters using agentic AI and a large language model (LLM)-driven framework.
Updated version of DeepSeek-R1 with enhanced reasoning, coding, math, and reduced hallucination.
Multi-modal vision-language model that understands text/img and creates informative responses
Detect and prevent sophisticated fraudulent activities for financial services with high accuracy.
State-of-the-art model for Polish language processing tasks such as text generation, Q&A, and chatbots.
State-of-the-art open model for reasoning, code, math, and tool calling - suitable for edge agents
Expressive and engaging text-to-speech, generated from a short audio sample.
Advanced reasoing MOE mode excelling at reasoning, multilingual tasks, and instruction following
FLUX.1-schnell is a distilled image generation model, producing high quality images at fast speeds
Design, test, and optimize a new generation of intelligence manufacturing data centers using digital twins.
State-of-the-art, multilingual model tailored to all 24 official European Union languages.
SOTA LLM pre-trained for instruction following and proficiency in Indonesian language and its dialects.
Efficient multimodal model excelling at multilingual tasks, image understanding, and fast-responses
High accuracy and optimized performance for transcription in 25 languages
Create high quality images using Flux.1 in ComfyUI, guided by 3D.
Superior inference efficiency with highest accuracy for scientific and complex math reasoning, coding, tool calling, and instruction following.
Build a custom deep researcher powered by state-of-the-art models that continuously process and synthesize multimodal enterprise data, enabling reasoning, planning, and refinement to generate comprehensive reports.
Develop AI powered weather analysis and forecasting application visualizing multi-layered geospatial data.
Investigate, understand, and interpret single cell data in minutes, not days by leveraging RAPIDS-singlecell, powered by NVIDIA RAPIDS
Easily run essential genomics workflows to save time leveraging Parabricks
Run computational-fluid dynamics (CFD) simulations
Generate exponentially large amounts of synthetic motion trajectories for robot manipulation from just a few human demonstrations.
Generalist model to generate future world state as videos from text and image prompts to create synthetic training data for robots and autonomous vehicles.
Generates future frames of a physics-aware world state based on simply an image or short video prompt for physical AI development.
Simulate, test, and optimize physical AI and robotic fleets at scale in industrial digital twins before real-world deployment.
End-to-end autonomous driving stack integrating perception, prediction, and planning with sparse scene representations for efficiency and safety.
High efficiency model with leading accuracy for reasoning, tool calling, chat, and instruction following.
Leading reasoning and agentic AI accuracy model for PC and edge.
Natural and expressive voices in multiple languages. For voice agents and brand ambassadors.
The NV-EmbedCode model is a 7B Mistral-based embedding model optimized for code retrieval, supporting text, code, and hybrid queries.
Distilled version of Llama 3.1 8B using reasoning data generated by DeepSeek R1 for enhanced performance.
Model for object detection, fine-tuned to detect charts, tables, and titles in documents.
Model for object detection, fine-tuned to detect charts, tables, and titles in documents.
Model for object detection, fine-tuned to detect charts, tables, and titles in documents.
Cutting-edge vision-language model exceling in retrieving text and metadata from images.
Route LLM requests to the best model for the task at hand.
Distilled version of Qwen 2.5 32B using reasoning data generated by DeepSeek R1 for enhanced performance.
Distilled version of Qwen 2.5 14B using reasoning data generated by DeepSeek R1 for enhanced performance.
Distilled version of Qwen 2.5 7B using reasoning data generated by DeepSeek R1 for enhanced performance.
Lightweight multilingual LLM powering AI applications in latency bound, memory/compute constrained environments
This workflow shows how generative AI can generate DNA sequences that can be translated into proteins for bioengineering.
Robust Speech Recognition via Large-Scale Weak Supervision.
State-of-the-art, high-efficiency LLM excelling in reasoning, math, and coding.
Continuously extract, embed, and index multimodal data for fast, accurate semantic search. Built on world-class NeMo Retriever models, the RAG blueprint connects AI applications to multimodal enterprise data wherever it resides.
Topic control model to keep conversations focused on approved topics, avoiding inappropriate content.
Industry leading jailbreak classification model for protection from adversarial attempts
Leading content safety model for enhancing the safety and moderation capabilities of LLMs
NVIDIA DGX Cloud trained multilingual LLM designed for mission critical use cases in regulated industries including financial services, government, heavy industry
This blueprint shows how generative AI and accelerated NIM microservices can design protein binders smarter and faster.
Transform PDFs into AI podcasts for engaging on-the-go audio content.
Trace and evaluate AI Agents with Weights & Biases.
Automate voice AI agents with NVIDIA NIM microservices and Pipecat.
Automate research, and generate blogs with AI Agents using LlamaIndex and Llama3.3-70B NIM LLM.
Generate detailed, structured reports on any topic using LangGraph and Llama3.3 70B NIM.
Document your github repositories with AI Agents using CrewAI and Llama3.3 70B NIM.
Multi-modal vision-language model that understands text/img/video and creates informative responses
Multilingual and cross-lingual text question-answering retrieval with long context support and optimized data storage efficiency.
Fine-tuned reranking model for multilingual, cross-lingual text question-answering retrieval, with long context support.
Context-aware chart extraction that can detect 18 classes for chart basic elements, excluding plot elements.
Model for object detection, fine-tuned to detect charts, tables, and titles in documents.
Converts streamed audio to facial blendshapes for realtime lipsyncing and facial performances.
This NVIDIA Omniverse™ Blueprint demonstrates how commercial software vendors can create interactive digital twins.
FourCastNet predicts global atmospheric dynamics of various weather / climate variables.
Advanced AI model detects faces and identifies deep fake images.
Ingest massive volumes of live or archived videos and extract insights for summarization and interactive Q&A
Enhance and modify high-quality compositions using real-time rendering and generative AI output without affecting a hero product asset.
Create intelligent virtual assistants for customer service across every industry
A bilingual Hindi-English SLM for on-device inference, tailored specifically for Hindi Language.
Llama-3.1-Nemotron-70B-Instruct is a large language model customized by NVIDIA in order to improve the helpfulness of LLM generated responses.
Rapidly identify and mitigate container security vulnerabilities with generative AI.
Sovereign AI model trained on Japanese language that understands regional nuances.
Sovereign AI model trained on Japanese language that understands regional nuances.
Enhance speech by correcting common audio degradations to create studio quality speech output.
Leaderboard topping reward model supporting RLHF for better alignment with human preferences.
Unique language model that delivers an unmatched accuracy-efficiency performance.
Robust image classification model for detecting and managing AI-generated content.
Sovereign AI model finetuned on Traditional Mandarin and English data using the Llama-3 architecture.
Sovereign AI model trained on Japanese language that understands regional nuances.
This blueprint shows how generative AI and accelerated NIM microservices can design optimized small molecules smarter and faster.
Cutting-edge MOE based LLM designed to excel in a wide array of generative AI tasks.
Cutting-edge MOE based LLM designed to excel in a wide array of generative AI tasks.
Optimized SLM for on-device inference and fine-tuned for roleplay, RAG and function calling
State-of-the-art small language model delivering superior accuracy for chatbot, virtual assistants, and content generation.
Lightweight multilingual LLM powering AI applications in latency bound, memory/compute constrained environments
Grounding dino is an open vocabulary zero-shot object detection model.
Enable smooth global interactions in 36 languages.
Record-setting accuracy and performance for English transcription.
State-of-the-art accuracy and speed for English transcriptions.
Advanced small language generative AI model for edge applications
Estimate gaze angles of a person in a video and redirect to make it frontal.
Verify compatibility of OpenUSD assets with instant RTX render and rule-based validation.
Multilingual text reranking model.
English text embedding model for question-answering retrieval.
Multilingual text question-answering retrieval, transforming textual information into dense vector representations.
Advanced LLM to generate high-quality, context-aware responses for chatbots and search engines.
Advanced LLM to generate high-quality, context-aware responses for chatbots and search engines.
Generates high-quality numerical embeddings from text inputs.
Visual Changenet detects pixel-level change maps between two images and outputs a semantic change segmentation mask
EfficientDet-based object detection network to detect 100 specific retail objects from an input video.
LLM to represent and serve the linguistic and cultural diversity of Southeast Asia
An MOE LLM that follows instructions, completes requests, and generates creative text.
GPU-accelerated model optimized for providing a probability score that a given passage contains the information to answer a question.
This LLM follows instructions, completes requests, and generates creative text.
An MOE LLM that follows instructions, completes requests, and generates creative text.
