Scalable Prompt Routing via Fine-Grained Latent Task Discovery

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhang, Yunyi, Adeshina, Soji, Guan, Sheng, Ganesh, Ashwin, Han, Zhen, Ioannidis, Vassilis N., Rangwala, Huzefa, Karypis, George
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915882972741632
author Zhang, Yunyi
Adeshina, Soji
Guan, Sheng
Ganesh, Ashwin
Han, Zhen
Ioannidis, Vassilis N.
Rangwala, Huzefa
Karypis, George
author_facet Zhang, Yunyi
Adeshina, Soji
Guan, Sheng
Ganesh, Ashwin
Han, Zhen
Ioannidis, Vassilis N.
Rangwala, Huzefa
Karypis, George
contents Prompt routing dynamically selects the most appropriate large language model from a pool of candidates for each query, optimizing performance while managing costs. As model pools scale to include dozens of frontier models with narrow performance gaps, existing approaches face significant challenges: manually defined task taxonomies cannot capture fine-grained capability distinctions, while monolithic routers struggle to differentiate subtle differences across diverse tasks. We propose a two-stage routing architecture that addresses these limitations through automated fine-grained task discovery and task-aware quality estimation. Our first stage employs graph-based clustering to discover latent task types and trains a classifier to assign prompts to discovered tasks. The second stage uses a mixture-of-experts architecture with task-specific prediction heads for specialized quality estimates. At inference, we aggregate predictions from both stages to balance task-level stability with prompt-specific adaptability. Evaluated on 10 benchmarks with 11 frontier models, our method consistently outperforms existing baselines and surpasses the strongest individual model while incurring less than half its cost.
format Preprint
id arxiv_https___arxiv_org_abs_2603_19415
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Scalable Prompt Routing via Fine-Grained Latent Task Discovery
Zhang, Yunyi
Adeshina, Soji
Guan, Sheng
Ganesh, Ashwin
Han, Zhen
Ioannidis, Vassilis N.
Rangwala, Huzefa
Karypis, George
Computation and Language
Artificial Intelligence
Machine Learning
Prompt routing dynamically selects the most appropriate large language model from a pool of candidates for each query, optimizing performance while managing costs. As model pools scale to include dozens of frontier models with narrow performance gaps, existing approaches face significant challenges: manually defined task taxonomies cannot capture fine-grained capability distinctions, while monolithic routers struggle to differentiate subtle differences across diverse tasks. We propose a two-stage routing architecture that addresses these limitations through automated fine-grained task discovery and task-aware quality estimation. Our first stage employs graph-based clustering to discover latent task types and trains a classifier to assign prompts to discovered tasks. The second stage uses a mixture-of-experts architecture with task-specific prediction heads for specialized quality estimates. At inference, we aggregate predictions from both stages to balance task-level stability with prompt-specific adaptability. Evaluated on 10 benchmarks with 11 frontier models, our method consistently outperforms existing baselines and surpasses the strongest individual model while incurring less than half its cost.
title Scalable Prompt Routing via Fine-Grained Latent Task Discovery
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2603.19415