LongVPO: From Anchored Cues to Self-Reasoning for Long-Form Video Preference Optimization

Xi Chen (the University of Hong Kong, University of Hong Kong) · Jiaqi Li (Beijing Institute for General Artificial Intelligence) · Zhenpeng Huang (Nanjing University) · zihan jia · Xinhao Li (University of California, San Diego) · Desen Meng (nanjing university) · Lingxue Song (Tsinghua University, Tsinghua University) · Liang Li (Alibaba Group) · Limin Wang (Nanjing University)

direct preference optimizationdispreferred responseslarge language modellong-form video understandinglongvpomulti-segment reasoning queriespositional biaspreference triplesquestion-specificity filteringrecursive captioning pipelinescene-level metadatasynthetic examplesultra-long videosvision-language modelsvisual-similarity filtering

We present LongVPO, a novel two‑stage Direct Preference Optimization framework that enables short‑context vision‑language models to robustly understand ultra‑long videos without any long‑video annotations. In Stage 1, we synthesize preference triples by anchoring questions to individual short clips, interleaving them with distractors, and applying visual‑similarity and question‑specificity filtering to mitigate positional bias and ensure unambiguous supervision. We also approximate the reference model’s scoring over long contexts by evaluating only the anchor clip, reducing computational overhead. In Stage 2, we employ a recursive captioning pipeline on long videos to generate scene-level metadata, and then use a large language model to craft multi-segment reasoning queries and dispreferred responses, aligning the model's preferences through multi-segment reasoning tasks. With only 16K synthetic examples and no costly human labels, \model{} outperforms the state‑of‑the‑art open‑source models on multiple long‑video benchmarks, while maintaining strong short‑video performance (e.g., on MVBench), offering a scalable paradigm for efficient long‑form video understanding.