Benchmark
MTEB(kor, v2) - PairClassification
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.
- Models measured
- 11
- Scores recorded
- 11
- Sources
- 1
- IQ weight
- 1.10
Score distribution
- 65.9%
- Mean
- 66.2%
- Median
- 7.4%
- Std dev
- 57.1%
- 10th percentile
- 75.3%
- 90th percentile
- 51.8%
- Min
This benchmark has a weight of 1.10 for embedding models, measured over 11 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 8BPairClassification· MTEB Leaderboard data | 77.3% |
| 2 | Qwen3 Embedding 4BPairClassification· MTEB Leaderboard data | 75.3% |
| 3 | Jina Embeddings v5 Omni SmallPairClassification· MTEB Leaderboard data | 70.3% |
| 4 | Jina Embeddings v5 Omni NanoPairClassification· MTEB Leaderboard data | 70% |
| 5 | Qwen3 Embedding 0.6BPairClassification· MTEB Leaderboard data | 68.2% |
| 6 | E5 Mistral 7B InstructPairClassification· MTEB Leaderboard data | 66.2% |
| 7 | BGE multilingual gemma2PairClassification· MTEB Leaderboard data | 65.7% |
| 8 | Multilingual E5 Large InstructPairClassification· MTEB Leaderboard data | 65.4% |
| 9 | snowflake-arctic-embed-l-v2.0PairClassification· MTEB Leaderboard data | 57.6% |
| 10 | Granite Embedding 278m MultilingualPairClassification· MTEB Leaderboard data | 57.1% |
| 11 | All MiniLM L6 V2PairClassification· MTEB Leaderboard data | 51.8% |