Skip to content

Benchmark

MTEB(eng, v2) - Retrieval

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

Models measured
41
Scores recorded
41
Sources
1
IQ weight
0.93

Score distribution

54.2%
Mean
54.7%
Median
7.9%
Std dev
44.5%
10th percentile
62.8%
90th percentile
26.7%
Min
26.7%69.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.93 for embedding models, measured over 41 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3 Embedding 8BRetrieval· MTEB Leaderboard data69.4%
2Qwen3 Embedding 4BRetrieval· MTEB Leaderboard data68.5%
3Gemini Embedding 001Retrieval· MTEB Leaderboard data64.4%
4INF Retriever v1Retrieval· MTEB Leaderboard data64.1%
5NV-Embed-v2Retrieval· MTEB Leaderboard data62.8%
6Qwen3 Embedding 0.6BRetrieval· MTEB Leaderboard data61.8%
7Jina Embeddings v5 Omni SmallRetrieval· MTEB Leaderboard data60.1%
8Jina Embeddings v5 Text SmallRetrieval· MTEB Leaderboard data60.1%
9Text Embedding 004Retrieval· MTEB Leaderboard data59.1%
10Jina Embeddings v5 Omni NanoRetrieval· MTEB Leaderboard data58.8%
11Jina Embeddings v5 Text NanoRetrieval· MTEB Leaderboard data58.8%
12Text Embedding 005Retrieval· MTEB Leaderboard data58.8%
13snowflake-arctic-embed-l-v2.0Retrieval· MTEB Leaderboard data58.6%
14snowflake-arctic-embed-m-v2.0Retrieval· MTEB Leaderboard data58.4%
15Text Embedding 3 LargeRetrieval· MTEB Leaderboard data58%
16E5 Mistral 7B InstructRetrieval· MTEB Leaderboard data57.6%
17Jina Embeddings v4Retrieval· MTEB Leaderboard data56.2%
18UAE-Large-V1Retrieval· MTEB Leaderboard data55.9%
19BGE Large En V1.5Retrieval· MTEB Leaderboard data55.4%
20BGE Base En V1.5Retrieval· MTEB Leaderboard data54.8%
21Text Multilingual Embedding 002Retrieval· MTEB Leaderboard data54.7%
22Jina Embeddings v3Retrieval· MTEB Leaderboard data54.3%
23Granite Embedding Small English R2Retrieval· MTEB Leaderboard data53.9%
24bge-small-en-v1.5Retrieval· MTEB Leaderboard data53.9%
25Text Embedding 3 SmallRetrieval· MTEB Leaderboard data53.5%
26Multilingual E5 Large InstructRetrieval· MTEB Leaderboard data53.5%
27Gte LargeRetrieval· MTEB Leaderboard data53.3%
28Gte BaseRetrieval· MTEB Leaderboard data51.9%
29Multilingual E5 LargeRetrieval· MTEB Leaderboard data51.5%
30Granite Embedding 278m MultilingualRetrieval· MTEB Leaderboard data51.5%
31embeddinggemma-2Retrieval· MTEB Leaderboard data51.3%
32nomic-embed-text-v1Retrieval· MTEB Leaderboard data50.7%
33E5 Base V2Retrieval· MTEB Leaderboard data49.7%
34E5 Large V2Retrieval· MTEB Leaderboard data49.3%
35nomic-embed-text-v1.5Retrieval· MTEB Leaderboard data48%
36Jina Embedding B EN v1Retrieval· MTEB Leaderboard data46.4%
37All Mpnet Base V2Retrieval· MTEB Leaderboard data44.5%
38All MiniLM L12 V2Retrieval· MTEB Leaderboard data43.7%
39All MiniLM L6 V2Retrieval· MTEB Leaderboard data42.9%
40Paraphrase Multilingual MPNet v2Retrieval· MTEB Leaderboard data37.1%
41Jina Embeddings v2 Base ENRetrieval· MTEB Leaderboard data26.7%