Skip to content

Benchmark

Agentset Embedding nDCG@10 - MSMARCO

Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.

Models measured
15
Scores recorded
15
Sources
1
IQ weight
0.00

Score distribution

94.1%
Mean
94.2%
Median
0.872%
Std dev
93.1%
10th percentile
95.4%
90th percentile
91.8%
Min
91.8%95.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.00 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 SmallMSMARCO· Agentset95.4%
2Qwen3 Embedding 4BMSMARCO· Agentset95.4%
3Text Embedding 3 LargeMSMARCO· Agentset94.7%
4Text Embedding 3 SmallMSMARCO· Agentset94.6%
5Jina Embeddings v3MSMARCO· Agentset94.4%
6voyage-3.5MSMARCO· Agentset94.4%
7voyage-3.5-liteMSMARCO· Agentset94.4%
8voyage-3-largeMSMARCO· Agentset94.2%
9BGE M3MSMARCO· Agentset94.1%
10Embed Multilingual V3.0MSMARCO· Agentset94.1%
11Gemini Embedding 2MSMARCO· Agentset93.9%
12Qwen3 Embedding 8BMSMARCO· Agentset93.7%
13Qwen3 Embedding 0.6BMSMARCO· Agentset93.3%
14voyage-4MSMARCO· Agentset93.1%
15Text Embedding 004MSMARCO· Agentset91.8%