Dynamic Model Routing and Cascading for Efficient LLM Inference: A Survey

The rapid growth of large language models (LLMs) with diverse capabilities, costs, and domains has created a critical need for intelligent model selection at inference time. While smaller models suffice for routine queries, complex tasks demand more capable models. However, static model deployment does not account for the complexity and domain of incoming queries, leading to suboptimal performance and increased costs. Dynamic routing systems that adaptively select models based on query characteristics have emerged as a solution to this challenge. This survey provides a systematic analysis of state-of-the-art multi-LLM routing approaches. Unlike mixture-of-experts architectures that route within a single model, this survey focuses on routing across multiple independently trained LLMs. We cover diverse routing paradigms, including query difficulty, human preferences, clustering, uncertainty quantification, reinforcement learning, multimodality, and cascading. For each paradigm, we analyze representative methods and examine key trade-offs. Our analysis reveals that effective multi-LLM routing requires balancing competing objectives.~Choosing the optimal routing strategy depends on deployment and computational constraints.~Well-designed routing systems can outperform even the most powerful individual models by strategically leveraging specialized capabilities across models while maximizing efficiency gains. Meanwhile, significant challenges remain in developing routing mechanisms that generalize across diverse architectures, modalities, and application scenarios.