Build Reasoning Models to Achieve Advanced Agentic AI Autonomy

Reasoning models represent a paradigm shift in AI for developing truly intelligent agentic systems. Unlike previous types of models, reasoning models can not only imitate human behavior, but can explore different ways of finding solutions on their own. The reasoning model’s approach also often leads to superior answer quality and often surprising and unexpected insights and solutions.

We'll explain how reasoning models like DeepSeek-R1 are built using various techniques such as distillation and reinforcement learning with scalable (non-human) feedback, and the applications these models are unlocking. We'll also cover open tools, like NVIDIA NeMo, that you can use to create, fine-tune, or distill reasoning models and customize the open models for your domain.