arXiv cs.CL August 13, 2026 · Papers

LLM Router: Rethinking Routing with Prefill Activations

arXiv:2603.20895v3 Announce Type: replace Abstract: Existing routers rely on semantic query features or handcrafted features, which often fail to capture model-specific failures or intrinsic task difficulty. We instead route using internal LLM activations, specifically the residual stream. Our key idea, Encoder-Target