Recent years have witnessed the rise of increasingly larger and more sophisticated language models (LMs) capable of performing every task imaginable, sometimes at (super)human level. In this talk, I will argue that in many realistic scenarios, solely relying on a single general-purpose LLM is suboptimal. A single LLM is likely to under-represent real-world data distributions, heterogeneous skills, and task-specific requirements. Instead, I will discuss Multi-LLM collaboration as an alternative to monolithic generative modeling. By orchestrating multiple LLMs, each with distinct roles, perspectives, or competencies, we can achieve more effective problem-solving while being more inclusive and explainable. I will illustrate this approach through two case studies: narrative story generation and visual question answering, showing how a society of agents can collectively tackle complex tasks while pursuing complementary subgoals. Additionally, I will explore how these agent societies leverage reasoning to improve performance.