DeepVoyager-VL: Incentivizing Vision-in-the-Loop Search for Long-Horizon Multimodal Agents

Hugging Face Daily Papers Papers

Summary

DeepVoyager-VL proposes a long-horizon multimodal deep-search framework that integrates visual evidence into intermediate reasoning, using a multimodal event graph for data synthesis and fine-tuning without reinforcement learning, achieving strong performance across ten benchmarks.

Multimodal large language models (MLLMs) have advanced visual understanding and reasoning, yet their static parametric knowledge limits their ability to address knowledge-intensive and dynamically evolving open-world problems. To move beyond this limitation, multimodal deep search has emerged as a key direction for open-world information access, evolving from single-turn factual retrieval toward long-horizon, multi-turn search guided by visual evidence. However, existing methods typically confine vision to the input or answer stage, overlooking its role in intermediate reasoning, and lack designs tailored to long-horizon interaction. Consequently, visual evidence rarely drives continued retrieval, constraining both interaction depth and reasoning span. To address these limitations, we propose DeepVoyager-VL, a long-horizon multimodal deep-search framework for vision-in-the-loop search. Specifically, we construct a multimodal event graph to drive data synthesis, yielding problems with intermediate visual dependencies and long reasoning chains. We then design an agent framework for active visual acquisition and on-demand image loading. Finally, we fine-tune models on the synthesized data without reinforcement learning. Extensive experiments across ten multimodal search benchmarks demonstrate the effectiveness of our method.
Original Article
View Cached Full Text

Cached at: 08/04/26, 09:38 AM

Paper page - DeepVoyager-VL: Incentivizing Vision-in-the-Loop Search for Long-Horizon Multimodal Agents

Source: https://huggingface.co/papers/2608.01827 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

Multimodallargelanguagemodels(MLLMs)haveadvancedvisualunderstandingandreasoning,yettheirstaticparametricknowledgelimitstheirabilitytoaddressknowledge-intensiveanddynamicallyevolvingopen-worldproblems.Tomovebeyondthislimitation,multimodaldeepsearchhasemergedasakeydirectionforopen-worldinformationaccess,evolvingfromsingle-turnfactualretrievaltowardlong-horizon,multi-turnsearchguidedbyvisualevidence.However,existingmethodstypicallyconfinevisiontotheinputoranswerstage,overlookingitsroleinintermediatereasoning,andlackdesignstailoredtolong-horizoninteraction.Consequently,visualevidencerarelydrivescontinuedretrieval,constrainingbothinteractiondepthandreasoningspan.Toaddresstheselimitations,weproposeDeepVoyager-VL,along-horizonmultimodaldeep-searchframeworkforvision-in-the-loopsearch.Specifically,weconstructamultimodaleventgraphtodrivedatasynthesis,yieldingproblemswithintermediatevisualdependenciesandlongreasoningchains.Wethendesignanagentframeworkforactivevisualacquisitionandon-demandimageloading.Finally,wefine-tunemodelsonthesynthesizeddatawithoutreinforcementlearning.Extensiveexperimentsacrosstenmultimodalsearchbenchmarksdemonstratetheeffectivenessofourmethod.

View arXiv pageView PDFProject pageGitHub1Add to collection

Get this paper in your agent:

hf papers read 2608\.01827

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.01827 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.01827 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.01827 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

VLX-VR: An Agentic-Aware Video Reasoning Model

arXiv cs.CL

VLX-VR is an agentic-aware video reasoning model that uses a Think–Memory–Observation loop and reinforcement learning to adaptively gather evidence, achieving state-of-the-art performance on the MINERVA benchmark.