Abstract:As LLMs proliferate with diverse capabilities and costs, LLM routing has emerged by learning to predict each LLM's quality and cost for a given query, then selecting the one with high quality and low cost. However, existing routers implicitly assume a single fixed quality and cost per LLM for each query, ignoring that the same LLM's quality varies with its output length. This causes routers to exclude powerful LLMs when their estimated cost exceeds the budget, missing the opportunity that these LLMs could still deliver high quality at reduced cost with shorter outputs. To address this, we introduce R2-Router, which treats output length budget as a controllable variable and jointly selects the best LLM and length budget, enforcing the budget via length-constrained instructions. This enables R2-Router to discover that a powerful LLM with constrained output can outperform a weaker LLM at comparable cost-efficient configurations invisible to prior methods. Together with the router framework, we construct R2-Bench, the first routing dataset capturing LLM behavior across diverse output length budgets. Experiments show that R2-Router achieves state-of-the-art performance at 4-5\times lower cost compared with existing routers. This work opens a new direction: routing as reasoning, where routers evolve from reactive selectors to deliberate reasoners that explore which LLM to use and at what cost budget. The code is publicly available at this https URL.
| Comments: | Accepted by ICML 2026 |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2602.02823 [cs.CL] |
| (or arXiv:2602.02823v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2602.02823 arXiv-issued DOI via DataCite |
Submission history
From: Jiaqi Xue [view email]
[v1]
Mon, 2 Feb 2026 21:23:51 UTC (1,199 KB)
[v2]
Fri, 29 May 2026 19:48:02 UTC (1,202 KB)