LLM Routing as Reasoning: A MaxSAT View
This addresses the challenge of efficient LLM selection for users with natural language preferences, though it is incremental as it applies an existing constraint optimization method to a new routing context.
The paper tackled the problem of routing queries to appropriate LLMs when user preferences are in natural language and model attributes are partially observable, by formulating it as a weighted MaxSAT/MaxSMT problem, and empirical results on a 25-model benchmark showed that language feedback yields near-feasible recommendation sets.
Routing a query through an appropriate LLM is challenging, particularly when user preferences are expressed in natural language and model attributes are only partially observable. We propose a constraint-based interpretation of language-conditioned LLM routing, formulating it as a weighted MaxSAT/MaxSMT problem in which natural language feedback induces hard and soft constraints over model attributes. Under this view, routing corresponds to selecting models that approximately maximize satisfaction of feedback-conditioned clauses. Empirical analysis on a 25-model benchmark shows that language feedback produces near-feasible recommendation sets, while no-feedback scenarios reveal systematic priors. Our results suggest that LLM routing can be understood as structured constraint optimization under language-conditioned preferences.