Benchmark
Agentset Embedding Win Rate - DBPedia
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages. Higher is better.
- Models measured
- 15
- Scores recorded
- 15
- Sources
- 1
- IQ weight
- 0.65
Score distribution
- 46.6%
- Mean
- 46.9%
- Median
- 15.3%
- Std dev
- 20%
- 10th percentile
- 61.9%
- 90th percentile
- 15%
- Min
This benchmark has a weight of 0.65 for embedding models, measured over 15 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 | voyage-3-largeDBPedia· Agentset | 73.3% |
| 2 | voyage-3.5-liteDBPedia· Agentset | 61.9% |
| 3 | voyage-4DBPedia· Agentset | 60% |
| 4 | Gemini Embedding 2DBPedia· Agentset | 59.4% |
| 5 | Qwen3 Embedding 8BDBPedia· Agentset | 58.1% |
| 6 | BGE M3DBPedia· Agentset | 50.6% |
| 7 | Text Embedding 3 LargeDBPedia· Agentset | 50% |
| 8 | Embed Multilingual V3.0DBPedia· Agentset | 46.9% |
| 9 | Qwen3 Embedding 4BDBPedia· Agentset | 45% |
| 10 | Text Embedding 3 SmallDBPedia· Agentset | 43.8% |
| 11 | Jina Embeddings v5 Text SmallDBPedia· Agentset | 43.3% |
| 12 | voyage-3.5DBPedia· Agentset | 42.5% |
| 13 | Jina Embeddings v3DBPedia· Agentset | 28.7% |
| 14 | Qwen3 Embedding 0.6BDBPedia· Agentset | 20% |
| 15 | Text Embedding 004DBPedia· Agentset | 15% |