← back to paper
arxiv: 2604.23321 · 2 revisions
MMEB-V3: Measuring the Performance Gaps of Omni-Modality Embedding Models