Nemotron Labs: How Open Models Give Enterprises and Nations AI They Can Trust, Control and Customize

Learn how NVIDIA Nemotron open models help enterprises build specialized AI they can trust, control and customize with accuracy, efficiency and flexibility. Editor's note: This post is part of the Nemotron Labs blog series, which explores how the latest open models, datasets and training techniques help businesses build specialized AI systems and applications on NVIDIA platforms. Each post highlights practical ways to use an open stack to deliver real value in production - from transparent research copilots to scalable AI agents.

Enterprises have plenty of powerful models to choose from. The real test is whether the AI an enterprise builds uniquely addresses the needs of the business: improving workflows, tapping into domain knowledge and exceeding standards for accuracy and trust. Increasingly, competitive AI advantage comes from how organizations build with available models, more than which one they choose.

Open models like NVIDIA Nemotron are built for customization - helping enterprises and nations build AI that's controllable, trustworthy and tailored to their needs. From Using AI to Owning Intelligence Specialized AI , such as autonomous agents and applications, are built with customized open models. These agents are built to do a defined task well, as the models used are tuned on proprietary knowledge and evaluated against real business outcomes.

That requires access to the model itself. Closed models advance what's possible and continue to push forward the frontier of general intelligence, but also set a ceiling on what enterprises can inspect, tune and improve. Open models remove that barrier - providing complete ownership and control.

The most effective agentic AI applications are systems of models where open models work alongside leading frontier models , each fulfilling the job it does best. High-performance reasoning models can handle complex planning while smaller models execute on specialized tasks. This lets enterprises right-size inference costs, improve accuracy on specific tasks and maintain flexibility as workflows evolve.

Customization Enterprises Can Trust Open models give enterprises something closed models cannot: full control to customize, inspect and improve AI against business needs. Public benchmarks measure general capability - but business-specific evaluation lets teams test against their own data, workflows and definition of accuracy - then improve from there. For example, the cost of a wrong answer is high for industries like healthcare and legal, where teams handle sensitive data and face strict accuracy requirements.

Organizations in these sectors must have visibility into how a model was trained, how it performs and the ability to improve it when necessary. With open models, teams can inspect their applications, run private evaluations against their own criteria and stand up reinforcement learning environments tuned to their own workflows. No routing of their proprietary data through a third party is required.

Companies across industries are already specializing Nemotron for their domains: Abridge is customizing Nemotron to build the first foundation model purpose-built for clinical conversations. Glean built Waldo , an agentic search model that pairs Nemotron with larger closed models to deliver enterprise search at significantly lower latency and with fewer tokens. H Company built Holotron 3 Nano by post-training Nemotron 3 Nano Omni on proprietary computer-use data, achieving higher than 76% accuracy on OSWorld-Verified - a benchmark on computer tasks - and matching other leading frontier models at a fraction of the cost.

Harvey post-trained Nemotron 3 Ultra on its legal benchmark and reached frontier-class accuracy - matching leading closed models on complex legal tasks at at least 10x lower cost per run . Heidi Health is delivering frontier-quality outcomes in clinical documentation without needing frontier-scale compute.