MVEB: Massive Video Embedding Benchmark
Provides a comprehensive, standardized benchmark for video embeddings, enabling fair comparison across diverse tasks and revealing modality interactions.
MVEB introduces a 23-task benchmark for video embeddings, evaluating 33 models and finding that no single model excels across all tasks; MLLM-based embeddings lead on most tasks, while multimodal binding excels on retrieval and zero-shot classification. Audio's contribution depends on annotation provenance, with a six-point gap.
We introduce the Massive Video Embedding Benchmark (MVEB), a 23-task benchmark for video embeddings spanning classification, zero-shot classification, clustering, pair classification, retrieval, and video-centric question answering. We evaluate 33 models and find that no single model dominates: MLLM-based embeddings lead on classification, clustering, pair classification, and QA; multimodal binding leads on retrieval and zero-shot classification; generative MLLMs without contrastive adaptation collapse on cross-modal tasks. Paired video-only vs. audio+video evaluations show that audio's contribution depends on dataset annotation provenance: audio helps when labels were produced from both modalities and hurts when they were produced from visuals alone, a six-point gap consistent across model families. MVEB is derived from MVEB+, a 184-task pool, and is designed to maintain task diversity while reducing evaluation cost. It integrates into the MTEB ecosystem for unified evaluation across text, image, audio, and video. We release MVEB and all 184 tasks along with code and a leaderboard at https://github.com/embeddings-benchmark/mteb.