NVIDIA-NeMo

Switchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model selection, benchmarking, and cost/performance optimization.

NVIDIA Object Oriented Agents: the Pythonic way to build AI Agents.

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

Evaluate and improve models and agents using environments

A scalable, agentic-first, and HuggingFace-native RL framework for research (9k lines).

Agentic RL on Any Harness at Scale

Open-source library for scalable, reproducible evaluation of AI models and benchmarks.

🕵️ NeMo Anonymizer: Detect and protect PII through context-aware replacement and rewriting

Make the agents you ship faster, more accurate, and safer.

State-of-the-art framework for fast, large-scale training and inference of diffusion models

:shield: NeMo Safe Synthesizer: Create private, safe versions of sensitive tabular datasets.

🎭 NeMo User Sim: Improve models and agents with population-grounded simulated users, even before production traffic exists.

NeMo SDG-PGMs is a user-friendly Python library for building probabilistic graphical models (PGMs) to generate synthetic data

A fully opensource framework for building and evaluating voice agents with open models.

🔌 🎨 NeMo Data Designer Plugins

Research preview: Tools to efficiently find signals in large sets of agent traces and inform agent improvements.

Research preview: skills that help your coding assistant build evaluations for custom agents, assess the quality of existing suites, and find behavior-coverage gaps. Works today with NeMo Gym and Harbor.