Skip to content

Benchmark

Agentset Embedding Recall@10 - 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
0.16

Score distribution

12%
Mean
12.2%
Median
0.534%
Std dev
10.8%
10th percentile
12.3%
90th percentile
10.5%
Min
10.5%12.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.16 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

#ModelScore
1Jina Embeddings v5 Text SmallDBPedia· Agentset12.3%
2Qwen3 Embedding 8BDBPedia· Agentset12.3%
3Text Embedding 3 LargeDBPedia· Agentset12.3%
4Text Embedding 3 SmallDBPedia· Agentset12.3%
5voyage-3-largeDBPedia· Agentset12.3%
6BGE M3DBPedia· Agentset12.2%
7Embed Multilingual V3.0DBPedia· Agentset12.2%
8voyage-4DBPedia· Agentset12.2%
9Jina Embeddings v3DBPedia· Agentset12.1%
10voyage-3.5DBPedia· Agentset12.1%
11Gemini Embedding 2DBPedia· Agentset12%
12voyage-3.5-liteDBPedia· Agentset12%
13Qwen3 Embedding 4BDBPedia· Agentset11.9%
14Text Embedding 004DBPedia· Agentset10.8%
15Qwen3 Embedding 0.6BDBPedia· Agentset10.5%