Skip to content

Benchmark

Agentset Embedding Win Rate - business reports

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
0.37

Score distribution

47.6%
Mean
49.4%
Median
12.8%
Std dev
32.5%
10th percentile
64.1%
90th percentile
22.5%
Min
22.5%75%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.37 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 Smallbusiness reports· Agentset75%
2Gemini Embedding 2business reports· Agentset64.1%
3voyage-4business reports· Agentset58.9%
4Text Embedding 3 Largebusiness reports· Agentset56.9%
5BGE M3business reports· Agentset51.2%
6Text Embedding 3 Smallbusiness reports· Agentset50%
7voyage-3.5business reports· Agentset50%
8Qwen3 Embedding 0.6Bbusiness reports· Agentset49.4%
9Embed Multilingual V3.0business reports· Agentset45.6%
10Qwen3 Embedding 4Bbusiness reports· Agentset45.6%
11voyage-3-largebusiness reports· Agentset41.7%
12Qwen3 Embedding 8Bbusiness reports· Agentset35.6%
13voyage-3.5-litebusiness reports· Agentset35%
14Text Embedding 004business reports· Agentset32.5%
15Jina Embeddings v3business reports· Agentset22.5%