Multimodal document question answering requires retrieving dispersed evidence from visually rich long documents and performing reliable reasoning over heterogeneous information. Existing multimodal RAG systems remain limited by two bottlenecks: static retrieval that ignores query complexity, and end-to-end Vision-Language Models (VLMs) that couple visual perception with logical reasoning, leading to inefficient computation and unstable answer generation. We propose AutoThinkRAG, a complexity-aware inference architecture for multimodal document QA. It has two components: (1) a Query Complexity Router that analyzes query difficulty and structure to adaptively select retrieval and reasoning paths; and (2) a Perception--Reasoning Decoupling architecture that uses a lightweight VLM as a high-fidelity visual interpreter to convert query-relevant visual cues into textual representations, which are then passed to an LLM for logical reasoning and answer synthesis. This design improves both efficiency and robustness, especially on long-document and unanswerable queries. Experiments on DocBench and MMLongBench show that AutoThinkRAG achieves 82.13\% and 51.29\% overall accuracy, respectively, while reducing per-query token consumption by 18.9\% and monetary cost by 18.2\%. Further analyses show that the gains are most pronounced on complex queries requiring adaptive retrieval and multi-step reasoning.
@article{arxiv.2603.05551,
title = {AutothinkRAG: Complexity-Aware Control of Retrieval-Augmented Reasoning for Image-Text Interaction},
author = {Jiashu Yang and Chi Zhang and Abudukelimu Wuerkaixi and Xuxin Cheng and Cao Liu and Ke Zeng and Xu Jia and Xunliang Cai},
journal= {arXiv preprint arXiv:2603.05551},
year = {2026}
}