UniScale: Adaptive Unified Inference Scaling via Online Joint Optimization of Model Routing and Test-Time Scaling
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917547352260608 |
|---|---|
| author | Huang, Kaiyu Wang, Xingyu Kong, Mingze Shi, Zhubo Hou, Yuqian Xu, Hong Dai, Zhongxiang Yu, Minchen Shi, Qingjiang |
| author_facet | Huang, Kaiyu Wang, Xingyu Kong, Mingze Shi, Zhubo Hou, Yuqian Xu, Hong Dai, Zhongxiang Yu, Minchen Shi, Qingjiang |
| contents | In real-world deployments of large language models (LLMs), balancing inference quality and computational cost has become a central challenge. Existing approaches tackle this trade-off along two largely independent dimensions: model routing, which switches among models of different scales to match request complexity, and test-time scaling (TTS), which adjusts inference-time compute within a fixed model for fine-grained control. However, this decoupled design introduces inherent limitations. Model routing yields coarse-grained, discrete performance changes due to the sparse set of model scales, while single-model TTS often encounters capacity ceilings and exhibits diminishing returns as compute increases. Moreover, treating the two mechanisms separately restricts adaptability in dynamic inference environments. To overcome these limitations, we introduce Unified Inference Scaling (UIS), which unifies model routing and TTS in a single optimization space. Building on this formulation, we propose UniScale, an online framework that models adaptive UIS as a contextual multi-armed bandit problem and learns inference policies via LinUCB. The framework incorporates efficiency-aware learning and cost modeling to ensure stable and scalable optimization over high-dimensional action spaces. Evaluation shows that UniScale effectively exploits the synergy in the UIS space to deliver a fine-grained and consistently better quality-cost trade-off across diverse, dynamic inference scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_30898 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | UniScale: Adaptive Unified Inference Scaling via Online Joint Optimization of Model Routing and Test-Time Scaling Huang, Kaiyu Wang, Xingyu Kong, Mingze Shi, Zhubo Hou, Yuqian Xu, Hong Dai, Zhongxiang Yu, Minchen Shi, Qingjiang Artificial Intelligence Computation and Language In real-world deployments of large language models (LLMs), balancing inference quality and computational cost has become a central challenge. Existing approaches tackle this trade-off along two largely independent dimensions: model routing, which switches among models of different scales to match request complexity, and test-time scaling (TTS), which adjusts inference-time compute within a fixed model for fine-grained control. However, this decoupled design introduces inherent limitations. Model routing yields coarse-grained, discrete performance changes due to the sparse set of model scales, while single-model TTS often encounters capacity ceilings and exhibits diminishing returns as compute increases. Moreover, treating the two mechanisms separately restricts adaptability in dynamic inference environments. To overcome these limitations, we introduce Unified Inference Scaling (UIS), which unifies model routing and TTS in a single optimization space. Building on this formulation, we propose UniScale, an online framework that models adaptive UIS as a contextual multi-armed bandit problem and learns inference policies via LinUCB. The framework incorporates efficiency-aware learning and cost modeling to ensure stable and scalable optimization over high-dimensional action spaces. Evaluation shows that UniScale effectively exploits the synergy in the UIS space to deliver a fine-grained and consistently better quality-cost trade-off across diverse, dynamic inference scenarios. |
| title | UniScale: Adaptive Unified Inference Scaling via Online Joint Optimization of Model Routing and Test-Time Scaling |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2605.30898 |