Nemotron
Developer Asset Hub for NVIDIA Nemotron — A one-stop resource for training recipes, usage cookbooks, datasets, and full end-to-end reference examples to build w
What is Nemotron?
Developer Asset Hub for NVIDIA Nemotron — A one-stop resource for training recipes, usage cookbooks, datasets, and full end-to-end reference examples to build with Nemotron models
Overview of Nemotron
What it is
NVIDIA Nemotron is a family of open, high-efficiency multimodal models purpose-built for agentic AI. The repository serves as a developer asset hub providing training recipes, usage cookbooks, datasets, and end-to-end reference examples for building with Nemotron models. The models are designed for coding, math, scientific reasoning, tool calling, instruction following, and visual reasoning. They can be deployed across edge, single GPU, or data center environments with support for NeMo, TensorRT-LLM, vLLM, SGLang, and NIM microservices.
Key features
- Open models with fully transparent training data, techniques, and weights - Compute efficiency through model pruning and optimization enabling higher throughput via TensorRT-LLM - High accuracy built on frontier open models with human-aligned reasoning for agentic workflows - Flexible deployment: edge, single GPU, or data center with NIM microservices - Modular building blocks (Nemotron Steps) for training, evaluation, synthetic data generation, and more - Reproducible training pipelines from raw data to deployment-ready models
Use cases
- Building agentic AI workflows that require long-running agents and high-volume task execution - Coding, math, scientific reasoning, tool calling, instruction following, and visual reasoning - Deploying multimodal perception sub-agents (e.g., Nemotron 3 Nano Omni with text, image, video, audio support)
- Training, fine-tuning, or understanding how a model was built using complete pipelines - Building end-to-end applications such as RAG, agents, and tool use
Getting started
The repository includes a Claude Code plugin called nemotron-customize. Install it with: text /plugin marketplace add NVIDIA/Nemotron /plugin install nemotron-customize@nvidia-nemotron Then start Claude Code from the repo root and invoke the skill: bash cd /path/to/Nemotron claude /nemotron-customize For general usage, see the repository README for detailed instructions on using Nemotron Steps, training recipes, and cookbooks.
License and pricing
The project is licensed under Apache-2.0 and is open source.
Based on the project's public README and repository metadata.
