Metraipolis › archive › dc_91c878a3a1c2

Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent — arXiv:2608.03979

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.03979
License (as archived)
CC0-1.0
Archived
2026-10-31T12:19:43.179Z
SHA-256 of the archived text
48d10fbf20b2df07e9046f7fd5a31c6c4bbc8d8a04add95697b76d2535c1ddf0
Stable URL of this page
https://metraipolis.com/documento/dc_91c878a3a1c2

How to cite

Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent — arXiv:2608.03979 — arXiv (en), https://arxiv.org/abs/2608.03979 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_91c878a3a1c2 (Metraipolis archive dc_91c878a3a1c2, sha256 48d10fbf20b2df07e9046f7fd5a31c6c4bbc8d8a04add95697b76d2535c1ddf0)
BibTeX
@misc{dc91c878a3a1c2,
  title = {Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent — arXiv:2608.03979},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.03979},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_91c878a3a1c2; sha256 48d10fbf20b2df07e9046f7fd5a31c6c4bbc8d8a04add95697b76d2535c1ddf0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03979. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We introduce Video-DeepResearch (Video-DR), extending multimodal agents from static images to continuous video streams, a setting that demands dense spatiotemporal grounding coupled with open-web exploration. Preliminary evaluations reveal two critical bottlenecks in current models: (1) modality bias, where agents bypass visual tools in favor of textual search, and (2) parametric knowledge leakage, where models rely on internal memory rather than genuine tool-augmented execution. To address these challenges, we propose Video-DR, featuring a decoupled perception-exploration pipeline with stage-wise tool unlocking that compels exhaustive cross-frame visual grounding prior to web retrieval. Our framework adopts a two-stage training recipe: supervised fine-tuning followed by Group Relative Policy Optimization (GRPO), enabling autonomous exploration that breaks the imitation-learning ceiling. Furthermore, we curate Video-DR-Bench, a human-AI collaborative benchmark comprising 200 complex, multi-hop VQA instances. Empirical results demonstrate that our Video-DeepResearch-35B-A3B establishes a new state-of-the-art of 64.0% average accuracy, surpassing proprietary Claude-4.5-Sonnet (59.0%) by 5.0 points and significantly outperforming GPT-5 (52.5%) and Gemini 2.5 Pro (57.5%). The 30B-A3B variant achieves 59.3%, competitive with Claude-4.5-Sonnet and demonstrating the effectiveness of our training paradigm even at compact scale. Code: this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.03979 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T12:05:53.105Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).