Benchmark
MTEB(Europe, v1)
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.
- Models measured
- 16
- Scores recorded
- 16
- Sources
- 1
- IQ weight
- 0.76
Score distribution
- 55%
- Mean
- 54.4%
- Median
- 8.3%
- Std dev
- 46.3%
- 10th percentile
- 69%
- 90th percentile
- 43.5%
- Min
This benchmark has a weight of 0.76 for embedding models, measured over 16 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Qwen3 Embedding 8B· MTEB Leaderboard data | 69.9% |
| 2 | Qwen3 Embedding 4B· MTEB Leaderboard data | 69% |
| 3 | Qwen3 Embedding 0.6B· MTEB Leaderboard data | 63.7% |
| 4 | Multilingual E5 Large Instruct· MTEB Leaderboard data | 62.2% |
| 5 | E5 Mistral 7B Instruct· MTEB Leaderboard data | 61.3% |
| 6 | Jina Embeddings v3· MTEB Leaderboard data | 60.7% |
| 7 | snowflake-arctic-embed-l-v2.0· MTEB Leaderboard data | 58.4% |
| 8 | Paraphrase Multilingual MPNet v2· MTEB Leaderboard data | 54.4% |
| 9 | E5 Large V2· MTEB Leaderboard data | 50.4% |
| 10 | UAE-Large-V1· MTEB Leaderboard data | 49.5% |
| 11 | E5 Base V2· MTEB Leaderboard data | 49.4% |
| 12 | BGE Large En V1.5· MTEB Leaderboard data | 48.5% |
| 13 | BGE Base En V1.5· MTEB Leaderboard data | 46.9% |
| 14 | nomic-embed-text-v1.5· MTEB Leaderboard data | 46.3% |
| 15 | bge-small-en-v1.5· MTEB Leaderboard data | 46.1% |
| 16 | All MiniLM L6 V2· MTEB Leaderboard data | 43.5% |