MTEB logo MTEB
MVEB Video-Text icon

MVEB Video-Text

Cite this benchmark

@inproceedings{anonymous2026mveb,
  author = {Adnan El Assadi and Roman Solomatin and Isaac Chung and Chenghao Xiao and Deep Shah and Shriya Sudhakar and Ayush Sunil Munot and M{\'a}rton Kardos and Niklas Muennighoff and Kenneth Enevoldsen},
  booktitle = {The 2026 Conference on Empirical Methods in Natural Language Processing},
  title = {{MVEB}: Massive Video Embedding Benchmark},
  url = {https://openreview.net/forum?id=p5KyLtjRPo},
  year = {2026},
}

Text and video embedding quality across retrieval, classification, clustering, pair classification, zero-shot classification, and video-centric QA, for models without an audio encoder.

Languages 16
Tasks 19
Task Types 6
Models 0