Benchmark
MTEB(Medical, v1)
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.
- Models measured
- 20
- Scores recorded
- 20
- Sources
- 1
- IQ weight
- 0.78
Score distribution
- 47.2%
- Mean
- 42.1%
- Median
- 13%
- Std dev
- 33.1%
- 10th percentile
- 61.3%
- 90th percentile
- 21.1%
- Min
This benchmark has a weight of 0.78 for embedding models, measured over 20 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 | 69.6% |
| 2 | Qwen3 Embedding 4B· MTEB Leaderboard data | 67.5% |
| 3 | Qwen3 Embedding 0.6B· MTEB Leaderboard data | 61.3% |
| 4 | Text Embedding 3 Large· MTEB Leaderboard data | 61% |
| 5 | snowflake-arctic-embed-l-v2.0· MTEB Leaderboard data | 59.4% |
| 6 | Multilingual E5 Large Instruct· MTEB Leaderboard data | 58.3% |
| 7 | snowflake-arctic-embed-m-v2.0· MTEB Leaderboard data | 57.3% |
| 8 | Text Embedding 3 Small· MTEB Leaderboard data | 55.6% |
| 9 | Granite Embedding 278m Multilingual· MTEB Leaderboard data | 50.9% |
| 10 | Gte Large· MTEB Leaderboard data | 42.1% |
| 11 | UAE-Large-V1· MTEB Leaderboard data | 41.4% |
| 12 | E5 Large V2· MTEB Leaderboard data | 41.3% |
| 13 | nomic-embed-text-v1· MTEB Leaderboard data | 40.7% |
| 14 | Gte Base· MTEB Leaderboard data | 40.4% |
| 15 | E5 Base V2· MTEB Leaderboard data | 40.1% |
| 16 | nomic-embed-text-v1.5· MTEB Leaderboard data | 37.3% |
| 17 | All MiniLM L12 V2· MTEB Leaderboard data | 33.2% |
| 18 | Jina Embedding B EN v1· MTEB Leaderboard data | 33.1% |
| 19 | All Mpnet Base V2· MTEB Leaderboard data | 32.7% |
| 20 | Jina Embeddings v2 Base EN· MTEB Leaderboard data | 21.1% |