Benchmark
MVEB(text, video, v1) - Mean (TaskType)
Vision benchmarks measure how well models understand images. Scores are percentages.
- Models measured
- 4
- Scores recorded
- 4
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 38.3%
- Mean
- 44.4%
- Median
- 12.8%
- Std dev
- 18.9%
- 10th percentile
- 53.7%
- 90th percentile
- 18.9%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen3-VL Embedding 8BMean (TaskType)· MTEB Multimodal Boards data docs | 53.7% |
| 2 | Jina Embeddings v5 Omni SmallMean (TaskType)· MTEB Multimodal Boards data docs | 44.4% |
| 3 | Jina Embeddings v5 Omni NanoMean (TaskType)· MTEB Multimodal Boards data docs | 36.2% |
| 4 | Qwen2.5-Omni 7BMean (TaskType)· MTEB Multimodal Boards data docs | 18.9% |