Watching, Reasoning, and Searching: A Video Deep Research Benchmark on Open Web for Agentic Video Reasoning

TL;DR

VideoDR is a new benchmark for open-web video question answering that tests models' abilities in cross-frame extraction, web retrieval, and multi-hop reasoning, highlighting key challenges for future research.

Contribution

This paper introduces VideoDR, the first comprehensive benchmark for open-web video question answering involving cross-frame reasoning and web retrieval, with detailed evaluation of current models.

Findings

Agentic models' performance varies with their ability to maintain video anchors.

Goal drift and long-horizon consistency are major bottlenecks.

VideoDR reveals key challenges for next-generation video reasoning agents.

Abstract

In real-world video question answering scenarios, videos often provide only localized visual cues, while verifiable answers are distributed across the open web; models therefore need to jointly perform cross-frame clue extraction, iterative retrieval, and multi-hop reasoning-based verification. To bridge this gap, we construct the first video deep research benchmark, VideoDR. VideoDR centers on video-conditioned open-domain video question answering, requiring cross-frame visual anchor extraction, interactive web retrieval, and multi-hop reasoning over joint video-web evidence; through rigorous human annotation and quality control, we obtain high-quality video deep research samples spanning six semantic domains. We evaluate multiple closed-source and open-source multimodal large language models under both the Workflow and Agentic paradigms, and the results show that Agentic is not consistently superior to Workflow: its gains depend on a model's ability to maintain the initial video anchors over long retrieval chains. Further analysis indicates that goal drift and long-horizon consistency are the core bottlenecks. In sum, VideoDR provides a systematic benchmark for studying video agents in open-web settings and reveals the key challenges for next-generation video deep research agents.