IRT-Router: Effective and Interpretable Multi-LLM Routing via Item Response Theory

TL;DR

IRT-Router is a novel framework that uses Item Response Theory to effectively and interpretably route user queries to the most suitable large language models, balancing performance and cost.

Contribution

It introduces an IRT-based routing method that models LLM capabilities and query difficulty, providing both accurate predictions and interpretability.

Findings

Outperforms baseline methods in effectiveness and interpretability

Demonstrates strong performance in cold-start scenarios

Enhances online generalization with semantic similarity-based warm-up

Abstract

Large language models (LLMs) have demonstrated exceptional performance across a wide range of natural language tasks. However, selecting the optimal LLM to respond to a user query often necessitates a delicate balance between performance and cost. While powerful models deliver better results, they come at a high cost, whereas smaller models are more cost-effective but less capable. To address this trade-off, we propose IRT-Router, a multi-LLM routing framework that efficiently routes user queries to the most suitable LLM. Inspired by Item Response Theory (IRT), a psychological measurement methodology, IRT-Router explicitly models the relationship between LLM capabilities and user query attributes. This not only enables accurate prediction of response performance but also provides interpretable insights, such as LLM abilities and query difficulty. Additionally, we design an online query warm-up technique based on semantic similarity, further enhancing the online generalization capability of IRT-Router. Extensive experiments on 20 LLMs and 12 datasets demonstrate that IRT-Router outperforms most baseline methods in terms of effectiveness and interpretability. Its superior performance in cold-start scenarios further confirms the reliability and practicality of IRT-Router in real-world applications. Code is available at https://github.com/Mercidaiha/IRT-Router.