TSRouter: Dynamic Modality-Model Selection for Time Series Reasoning

2026-07-09

AI Summary

This work addresses the significant performance–cost trade-offs among heterogeneous models—such as large language models and vision-language models—in time series reasoning, where dynamic selection of the optimal model combination per query is crucial. The authors propose TSRouter, the first approach to introduce heterogeneous graph modeling into this setting, constructing a unified graph encompassing tasks, queries, modalities, and models. By leveraging graph neural networks to capture complex interactions within this structure, TSRouter reformulates model routing as a context-aware candidate scoring problem, enabling user-specified performance–cost trade-offs. The method operates in a zero-shot, plug-and-play manner and demonstrates strong generalization, achieving 16%–46% improvements over baselines across four time series reasoning tasks while maintaining high efficiency and low computational overhead.

Abstract

Time series reasoning is essential for real-world problem-solving. While both Large Language Models (LLMs) and Vision-Language Models (VLMs) can reason about time-series data, their capabilities are complementary: LLMs process time series as text sequences and thus preserve exact numerical understanding, but struggle with global patterns, whereas VLMs efficiently capture these patterns by visualizing time series but may lose fine-grained details. Moreover, models vary significantly in task-specific expertise and inference costs. Dynamically selecting the most suitable modality and model for each query is therefore crucial, yet challenging because it requires modeling the complex interactions among tasks, queries, modalities, and models, which carry rich contextual signals. To this end, we introduce TSRouter, a graph-based dynamic routing framework. TSRouter constructs a heterogeneous graph of task, query, modality, and model nodes to contextualize the interactions among query characteristics, modality attributes, and model capabilities. TSRouter formulates routing as a candidate scoring problem, where each modality-model pair is evaluated based on user-defined performance-cost preferences to select the optimal candidate. Comprehensive evaluations on 4 distinct time series reasoning tasks reveal that TSRouter substantially outperforms diverse baselines with 16\% to 46\% relative improvements. Furthermore, TSRouter demonstrates robust zero-shot plug-and-play generalization to unseen models and novel tasks and preserves high performance while reducing computational overhead through cost-aware optimization. Our code is available at https://github.com/tianyi-lab/TSRouter.

Problem

• time series reasoning
• modality selection
• model selection
• dynamic routing
• performance-cost trade-off

Innovation

• dynamic routing
• heterogeneous graph
• modality selection
• time series reasoning
• cost-aware optimization