# MTA-Agent: An Open Recipe for Multimodal Deep Search Agents Source: [MTA-Agent: An Open Recipe for Multimodal Deep Search Agents]() Collection index: [llms.txt]() Data updated: 2026-04-10T05:00:04.898Z Authors: Xiangyu Peng, Can Qin, An Yan, Xinyi Yang, Zeyuan Chen, Ran Xu, Chien-Sheng Wu Published: 2026-04-07T19:01:45.000Z ## Abstract Multimodal large language models (MLLMs) have demonstrated strong capabilities in visual understanding, yet they remain limited in complex, multi-step reasoning that requires deep searching and integrating visual evidence with external knowledge. In this work, we address this challenge by constructing high-quality, verified multi-hop vision-language training data for multimodal deep-search agents. We propose a Multi-hop Tool-Augmented Agent for Evidence-based QA Synthesis (MTA-Agent), which automatically selects tools and their parameters to retrieve and validate evidence from both visual and textual sources and generates structured multi-hop question-answer trajectories. Starting from diverse VQA seed datasets, our pipeline produces a large-scale training dataset, MTA-Vision-DeepSearch, containing 21K high-quality multi-hop examples. The data is filtered through a multi-stage verification process to ensure factual consistency and answer uniqueness. Using MTA-Vision-DeepSearch, a 32B open-source multimodal search agent achieves state-of-the-art performance, reaching an average of 54.63\\% across six challenging benchmarks, outperforming GPT-5 (51.86\\%), Gemini-2.5-Pro (50.98\\%), and Gemini-3-Pro (54.46\\%) under the same tool settings. We further show that training on our data improves both reasoning depth and tool-use behavior, increasing the average number of steps from 2.27 to 4.28, and leading to more systematic and persistent search strategies. Additionally, we demonstrate that training can be performed without real-time tool calls by replaying cached interactions, significantly reducing training cost. Importantly, we present MTA-Agent as a fully open recipe for multimodal deep search: we release the entire dataset, training trajectories, and implementation details to enable reproducibility and future research on open multimodal search agents. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Multimodal Deep Search on MTA (test)](): Multimodal Deep Search; MTA (test); Accuracy: 29.78. - [Multimodal Deep Search on BC-VL](): Multimodal Deep Search; BC-VL; Accuracy: 53.77. - [Multimodal Deep Search on MMSrch](): Multimodal Deep Search; MMSrch; Accuracy: 82.35. - [Multimodal Deep Search on FVQA (Accuracy)](): Multimodal Deep Search; FVQA; Accuracy: 76. - [Multimodal Deep Search on HR-MMSrch](): Multimodal Deep Search; HR-MMSrch; Accuracy: 53.95. - [Multimodal Deep Search on MMSrch-Plus](): Multimodal Deep Search; MMSrch-Plus; Accuracy: 31.93. ## Original sources and implementation [Original paper]() [PDF]()