How Do Thinking and Reasoning Models Work?

Explore the fundamentals of how large language models develop "thinking" and "reasoning" capabilities in this 13-minute explainer video from Google. Learn what actually happens when a model is "thinking" and discover the training methodologies that enhance LLM reasoning abilities. Understand key concepts including scaling laws, test-time compute, and reinforcement learning from verifiable rewards that enable these advanced AI capabilities. Gain insights into the technical mechanisms behind modern thinking models and how they differ from traditional language models in their approach to problem-solving and logical reasoning.