analiza aiqbench
Visual-Seeker advances the field of multimodal large language models by introducing a visual-native search paradigm that actively reasons over fine-grained image details rather than treating vision as static input. This approach enables robust multi-hop reasoning and factual grounding in complex, open-world scenarios, outperforming several proprietary models on key benchmarks. The work represents a significant step toward agentic systems capable of dynamic cross-modal evidence harvesting for deep web search.
Visual-Seeker: Towards Visual-Native Multimodal Agentic Search via Active Visual Reasoning
arXiv:2606.15231v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) have demonstrated impressive capabilities in many visual tasks, but they often struggle with factual grounding when confronted with complex, open-world scenarios. While recent multimodal deep search agents attempt to address this issue by utilizing external tools, the visual-native search paradigm remains underexplored. Existing methods primarily rely on simple images with explicit semantics and text-only evidence trajectories, limiting the agent's ability to perform multi-hop, cross-modal reasoning and search. To address these limitations, we propose Visual-Seeker, a visual-native multimodal deep search agent via active visual reasoning. Rather than treating vision as a static input, our agent actively attends to fine-grained visual details, dynamically harvests visual evidence throughout the search process. To unlock its visual-native potential, we design an active visual reasoning data pipeline and synthesize 5K high-quality multimodal trajectories for model training. Extensive experiments demonstrate the state-of-the-art performance across five challenging multimodal search benchmarks, even surpassing several proprietary models, validating robust visual-native reasoning and search in real-world web environments. The code and data can be accessed at: https://github.com/ZhengboZhang/Visual-Seeker.
Abstract:Multimodal large language models (MLLMs) have demonstrated impressive capabilities in many visual tasks, but they often struggle with factual grounding when confronted with complex, open-world scenarios. While recent multimodal deep search agents attempt to address this issue by utilizing external tools, the visual-native search paradigm remains underexplored. Existing methods primarily rely on simple images with explicit semantics and text-only evidence trajectories, limiting the agent's ability to perform multi-hop, cross-modal reasoning and search. To address these limitations, we propose Visual-Seeker, a visual-native multimodal deep search agent via active visual reasoning. Rather than treating vision as a static input, our agent actively attends to fine-grained visual details, dynamically harvests visual evidence throughout the search process. To unlock its visual-native potential, we design an active visual reasoning data pipeline and synthesize 5K high-quality multimodal trajectories for model training. Extensive experiments demonstrate the state-of-the-art performance across five challenging multimodal search benchmarks, even surpassing several proprietary models, validating robust visual-native reasoning and search in real-world web environments. The code and data can be accessed at: this https URL.
Artificial Intelligence (cs.AI)
arXiv:2606.15231 [cs.AI]
(or arXiv:2606.15231v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.15231 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 191 cuvinte
Subiecte conexe
[AINews] not much happened today
Latent Space · 10 sept. 2026
[AINews] High Taste Testers Ep. 1
Latent Space · 17 oct. 2026
Towards On-Policy Data Evolution for Visual-Native Multimodal Deep Search Agents
arXiv · 8 iun. 2026
[AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale
Latent Space · 12 sept. 2026