A Benchmark and Agentic Framework for Omni-Modal Reasoning and Tool Use in Long Videos
This work addresses the challenge of evaluating and improving multimodal large language models for real-world long-form video understanding, providing a practical benchmark for researchers, but it is incremental as it builds on existing benchmarking efforts.
The authors tackled the problem of long-form multimodal video understanding by introducing LongShOTBench, a diagnostic benchmark with open-ended questions and tool-use tasks, and LongShOTAgent, an agentic system for analysis. Results show state-of-the-art models like Gemini-2.5-Flash achieving 52.95% accuracy, open-source models below 30%, and LongShOTAgent at 44.66%, highlighting significant performance gaps.
Long-form multimodal video understanding requires integrating vision, speech, and ambient audio with coherent long-range reasoning. Existing benchmarks emphasize either temporal length or multimodal richness, but rarely both and while some incorporate open-ended questions and advanced metrics, they mostly rely on single-score accuracy, obscuring failure modes. We introduce LongShOTBench, a diagnostic benchmark with open-ended, intent-driven questions; single- and multi-turn dialogues; and tasks requiring multimodal reasoning and agentic tool use across video, audio, and speech. Each item includes a reference answer and graded rubric for interpretable, and traceable evaluation. LongShOTBench is produced via a scalable, human-validated pipeline to ensure coverage and reproducibility. All samples in our LongShOTBench are human-verified and corrected. Furthermore, we present LongShOTAgent, an agentic system that analyzes long videos via preprocessing, search, and iterative refinement. On LongShOTBench, state-of-the-art MLLMs show large gaps: Gemini-2.5-Flash achieves 52.95%, open-source models remain below 30%, and LongShOTAgent attains 44.66%. These results underscore the difficulty of real-world long-form video understanding. LongShOTBench provides a practical, reproducible foundation for evaluating and improving MLLMs. All resources are available on GitHub: https://github.com/mbzuai-oryx/longshot.