NVIDIA Launches Nemotron 3 Nano Omni Model, Unifying Vision, Audio and Language for up to 9x More Efficient AI Agents

<div id="bsf_rt_marker"></div><p>AI agent systems today juggle separate models for vision, speech and language — losing time and context as they pass data from one model to the other.</p> <p><span style="font-weight: 400;">Unveiled today, NVIDIA Nemotron 3 Nano Omni is an open multimodal model that brings these capabilities together into one system, </span><span style="font-weight: 400;">enabling agents to deliver faster, smarter responses with advanced reasoning across video, audio, image and text. </span><span style="font-weight: 400;">This best-in-class model gives enterprises and developers a production path for more efficient and accurate multimodal AI agents with full deployment flexibility and control. </span></p> <p><span style="font-weight: 400;">Nemotron 3 Nano Omni sets a new efficiency frontier for open multimodal models with leading accuracy and low cost, <a target="_blank" href="https://developer.nvidia.com/blog/nvidia-nemotron-3-nano-omni-powers-multimodal-agent-reasoning-in-a-single-efficient-open-model">topping six leaderboards</a> for complex document intelligence, and video and audio understanding.</span></p> <aside style="float: right; width: 320px; max-width: 100%; margin: 0.25rem 0 1.25rem 1.75rem; padding: 1.25rem 1.5rem 1.5rem; background: #fafafa; border-top: 3px solid #76b900; font-family: -apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,'Helvetica Neue',Arial,sans-serif; color: #1a1a1a; font-size: 15px; line-height: 1.55; box-sizing: border-box;"> <p style="margin: 0 0 0.875rem 0; font-size: 11px; letter-spacing: 0.12em; text-transform: uppercase; font-weight: bold; color: #76b900;">At a Glance</p> <div style="padding: 0 0 0.75rem 0;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">What it is</p> <p style="margin: 0; color: #1a1a1a;">An open, omni-modal reasoning model — the highest-efficiency open multimodal model of its kind with leading accuracy</p> </div> <div style="padding: 0.75rem 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">What it handles</p> <p style="margin: 0; color: #1a1a1a;">Text, images, audio, video, documents, charts and graphical interfaces (input); text (output)</p> </div> <div style="padding: 0.75rem 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">Who it’s for</p> <p style="margin: 0; color: #1a1a1a;">Enterprises and developers building fast and reliable, agentic systems that need a multimodal perception sub-agent</p> </div> <div style="padding: 0.75rem 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">How it works</p> <p style="margin: 0; color: #1a1a1a;">Functions as the “eyes and ears” in a system of agents, working alongside models like Nemotron 3 Super and Ultra or other proprietary models</p> </div> <div style="padding: 0.75rem 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">Why it matters</p> <p style="margin: 0; color: #1a1a1a;">Leading multimodal accuracy and 9x higher throughput than other open omni models with the same interactivity, resulting in lower cost and better scalability without sacrificing responsiveness.</p> </div> <div style="padding: 0.75rem 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">Architecture</p> <p style="margin: 0; color: #1a1a1a;">30B-A3B hybrid MoE with Conv3D, EVS, 256K context</p> </div> <div style="padding: 0.75rem 0 0 0; border-top: 1px solid #e5e5e5;"> <p style="margin: 0 0 0.2rem 0; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; font-weight: 600; color: #6b6b6b;">Availability</p> <p style="margin: 0; color: #1a1a1a;">April 28th, 2026 via Hugging Face, OpenRouter, build.nvidia.com and 25+ partner platforms</p> </div> </aside> <p>AI and<span style="font-weight: 400;"> software companies already adopting Nemotron 3 Nano Omni include </span><a target="_blank" href="https://www.aible.com/nemotron3-nano-omni-ai-agent"><span style="font-weight: 400;">Aible</span></a><span style="font-weight: 400;">, </span><a target="_blank" href="https://appliedscientific.ai/research/scientific-ai-literature-agent-nvidia-nemotron-nano-omni?utm_source=nvidia-blog"><span style="font-weight: 400;">Applied Scientific Intelligence (ASI)</span></a><span style="font-weight: 400;">, </span><a target="_blank" href="https://info.eka.care/services/how-ekacare-is-building-agentic-multimodal-healthcare-for-india-scale-patient-care-with-nvidia-nemotron-3-nano-omni"><span style="font-weight: 400;">Eka Care</span></a><span style="font-weight: 400;">, </span><span style="font-weight: 400;">Foxconn</span><span style="font-weight: 400;">, </span><span style="font-weight: 400;"><a target="_blank" href="https://hcompany.ai/holotron3">H Company</a>, Palantir and </span><a target="_blank" href="https://pyler.tech/articles/scaling-trustworthy-video-safety-with-nvidia-nemotron-3-nano-omni"><span style="font-weight: 400;">Pyler</span></a><span style="font-weight: 400;">,</span><span style="font-weight: 400;"> with </span><span style="font-weight: 400;">Dell Technologies</span><span style="font-weight: 400;">, </span><span style="font-weight: 400;">Docusign, Infosys, <a target="_blank" href="https://www.k-dense.ai/blog/nvidia-nemotron-nano-omni-multimodal-agentic-science">K-Dense</a>, Lila, Oracle </span><span style="font-weight: 400;">and </span><a target="_blank" href="https://zefr.com/press/zefr-evaluates-nvidia-nemotron-3-nano-omni-to-power-cognition-ai"><span style="font-weight: 400;">Zefr</span></a><span style="font-weight: 400;"> evaluating the model. </span></p> <p><span style="font-weight: 400;">“To build useful agents, you can’t wait seconds for a model to interpret a screen,”</span> <span style="font-weight: 400;">said Gautier Cloix, CEO of H Company.</span> <span style="font-weight: 400;">“By building on Nemotron 3 Nano Omni, our agents can rapidly interpret full HD screen recordings — something that wasn’t practical before. This isn’t just a speed boost: It’s a fundamental shift in how our agents perceive and interact with digital environments in real time.”</span></p> <h2><b>Nemotron 3 Nano Omni Enables Faster, Leaner Multimodal Agents</b></h2> <p><span style="font-weight: 400;">Consider an AI agent for customer support processing a screen recording while analyzing uploaded call audio and checking data logs — or an agent for finance tasked with parsing PDFs, spreadsheets, charts and voice notes. Today, most agentic systems accomplish these tasks with separate models for vision, speech and language. </span></p> <p><span style="font-weight: 400;">This approach increases latency through repeated inference passes, fragments context across modalities, and adds cost and inaccuracies over time.</span></p> <p><span style="font-weight: 400;">By combining vision and audio encoders within its 30B-A3B, hybrid </span><a target="_blank" href="https://www.nvidia.com/en-us/glossary/mixture-of-experts/"><span style="font-weight: 400;">mixture-of-experts</span></a><span style="font-weight: 400;"> architecture, Nemotron 3 Nano Omni eliminates the need for separate perception models, driving inference efficiency at scale. It pairs this efficiency with strong multimodal perception accuracy, enabling <a target="_blank" href="https://huggingface.co/blog/nvidia/nemotron-3-nano-omni-multimodal-inteligence">AI systems to achieve 9x higher throughput</a> than other open omni models with the same interactivity. The result is lower costs and better scalability without sacrificing responsiveness or quality. </span></p> <p><span style="font-weight: 400;">In agentic systems, Nemotron 3 Nano Omni can work alongside proprietary cloud models or other NVIDIA Nemotron open models — such as Nemotron 3 Super for high-frequency execution or Nemotron 3 Ultra for complex planning — as well as proprietary models from other providers, to power sub-agents for agentic workflows such as computer use, document intelligence and audio-video reasoning.</span></p> <ul> <li><b>Computer use agents —</b><span style="font-weight: 400;"> Nemotron 3 Nano Omni powers the perception loop for agents navigating graphical user interfaces, reasoning over onscreen content and understanding user interface state over time. </span><span style="font-weight: 400;">H Company’s latest </span><a target="_blank" href="https://www.youtube.com/watch?v=kSi9JS2l0Ww"><span style="font-weight: 400;">computer usage agent</span></a><span style="font-weight: 400;">, powered by Nemotron 3 Nano Omni, uses a native input resolution of 1920&#215;1080 pixels to achieve high-fidelity visual reasoning. In preliminary evaluations on the OSWorld benchmark, this integration showed a significant leap in navigating complex graphical interfaces and used Nemotron 3 Nano Omni’s ability to process very high-resolution images.</span><span style="font-weight: 400;"> </span></li> <li><b>Document intelligence</b><span style="font-weight: 400;"> — Interprets documents, charts, tables, screenshots and mixed-media inputs, enabling agents to reason across visual structure and text content coherently. Critical for enterprise analysis and compliance workflows.</span></li> <li><b>Audio and video understanding</b><span style="font-weight: 400;"> — For customer service, research and monitoring workflows, Nemotron 3 Nano Omni maintains audio-video context, tying what was said, shown and documented into a single reasoning stream instead of disconnected summaries.</span></li> </ul> <p><span style="font-weight: 400;"><img loading="lazy" decoding="async" class="aligncenter size-medium wp-image-92736" src="https://blogs.nvidia.com/wp-content/uploads/2026/04/nemotron-3-nano-omni-graphic-960x260.jpg" alt="" width="960" height="260" srcset="https://blogs.nvidia.com/wp-content/uploads/2026/04/nemotron-3-nano-omni-graphic-960x260.jpg 960w, https://blogs.nvidia.com/wp-content/uploads/2026/04/nemotron-3-nano-omni-graphic-630x171.jpg 630w, https://blogs.nvidia.com/wp-content/uploads/2026/04/nemotron-3-nano-omni-graphic.jpg 1026w" sizes="auto, (max-width: 960px) 100vw, 960px" /></span></p> <h2><b>Open and Customizable, Deployable Anywhere</b></h2> <p><span style="font-weight: 400;">Nemotron 3 Nano Omni is released with open weights, datasets and training techniques — giving organizations full transparency and control over how the model is customized and deployed. </span></p> <p><span style="font-weight: 400;">Developers can use tools like </span><a target="_blank" href="https://www.nvidia.com/en-us/ai-data-science/products/nemo/"><span style="font-weight: 400;">NVIDIA NeMo</span></a><span style="font-weight: 400;"> for customization, evaluation and optimization for domain-specific use cases. Because the Nemotron family of models is open, organizations can deploy them in environments that meet regulatory, sovereignty or data localization requirements. </span><span style="font-weight: 400;"><br /> </span><span style="font-weight: 400;"><br /> </span><span style="font-weight: 400;">The Nemotron 3 family — including Nano, Super and Ultra models — has seen over </span><span style="font-weight: 400;">50 million downloads in the past year</span><span style="font-weight: 400;">. Omni extends the family’s capabilities into multimodal and agentic domains. </span></p> <p><span style="font-weight: 400;">The model is available on </span><a target="_blank" href="https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16"><span style="font-weight: 400;">Hugging Face</span></a><span style="font-weight: 400;">, </span><a target="_blank" href="https://openrouter.ai/nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free"><span style="font-weight: 400;">OpenRouter</span></a> <span style="font-weight: 400;">and </span><a target="_blank" href="https://build.nvidia.com/nvidia/nemotron-3-nano-omni-30b-a3b-reasoning"><span style="font-weight: 400;">build.nvidia.com</span></a><span style="font-weight: 400;"> as an NVIDIA NIM microservice and through a broad ecosystem of </span><a target="_blank" href="https://www.nvidia.com/en-us/data-center/gpu-cloud-computing/partners/"><span style="font-weight: 400;">NVIDIA Cloud Partners</span></a><span style="font-weight: 400;">, inference platforms</span><span style="font-weight: 400;"> and cloud service providers. </span></p> <p><span style="font-weight: 400;">Its open, lightweight architecture supports consistent deployment from local systems like <a target="_blank" href="https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/">NVIDIA Jetson</a> hardware, </span><a target="_blank" href="https://www.nvidia.com/en-us/products/workstations/dgx-spark/"><span style="font-weight: 400;">NVIDIA DGX Spark</span></a><span style="font-weight: 400;"> and </span><a target="_blank" href="https://www.nvidia.com/en-us/products/workstations/dgx-station/"><span style="font-weight: 400;">DGX Station</span></a><span style="font-weight: 400;"> to data center and cloud environments. </span></p> <p><i><span style="font-weight: 400;">Visit the NVIDIA technical blog for </span></i><a target="_blank" href="https://developer.nvidia.com/blog/nvidia-nemotron-3-nano-omni-powers-multimodal-agent-reasoning-in-a-single-efficient-open-model"><i><span style="font-weight: 400;">tutorials, cookbooks and deployment guides</span></i></a><i> </i><i></i><i><span style="font-weight: 400;">for Nemotron 3 Nano Omni use cases. </span></i><i></i><i><span style="font-weight: 400;">S</span></i><i><span style="font-weight: 400;">tay up to date on agentic AI, </span></i><a target="_blank" href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/"><i><span style="font-weight: 400;">NVIDIA Nemotron</span></i></a><i><span style="font-weight: 400;"> and more by subscribing to </span></i><a target="_blank" href="https://www.nvidia.com/en-us/executive-insights/generative-ai-tools/?modal=stay-inf"><i><span style="font-weight: 400;">NVIDIA news</span></i></a><i><span style="font-weight: 400;">,</span></i><a target="_blank" href="https://developer.nvidia.com/community"><i><span style="font-weight: 400;"> joining the community</span></i></a><i><span style="font-weight: 400;"> and following NVIDIA AI on </span></i><a target="_blank" href="https://www.linkedin.com/showcase/nvidia-ai/posts/?feedView=all"><i><span style="font-weight: 400;">LinkedIn</span></i></a><i><span style="font-weight: 400;">, </span></i><a target="_blank" href="https://www.instagram.com/nvidiaai/?hl=en"><i><span style="font-weight: 400;">Instagram</span></i></a><i><span style="font-weight: 400;">, </span></i><a target="_blank" href="https://x.com/NVIDIAAIDev"><i><span style="font-weight: 400;">X</span></i></a><i><span style="font-weight: 400;"> and </span></i><a target="_blank" href="https://www.facebook.com/NVIDIAAI"><i><span style="font-weight: 400;">Facebook</span></i></a><i><span style="font-weight: 400;">.  </span></i></p> <p><i><span style="font-weight: 400;">Explore </span></i><a target="_blank" href="https://youtube.com/playlist?list=PL5B692fm6--vdRKB14FImVi7MTJ77zjn4&amp;feature=shared"><i><span style="font-weight: 400;">self-paced video tutorials and livestreams</span></i></a><i><span style="font-weight: 400;">.</span></i><i><span style="font-weight: 400;"><br /> </span></i></p>