MTEB Clustering

Runs mini-batch k-means over a model's embeddings and scores the clusters against ground-truth labels with V-measure, across 11 mostly-English datasets.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categoryembedding
Subcategoryclustering
Page statusactive
MetricV-measure
Directionhigher_is_better
Unit%
Dataset size11
Dataset licenceVaries by dataset.
PublisherHugging Face and Cohere (original authors); maintained today by the open embeddings-benchmark community

What it measures

A set of texts is embedded and a mini-batch k-means model is fit directly on those embeddings, with k set to the number of ground-truth categories. No labels are used during fitting; cluster assignments are then compared to the true category labels.

Task format

Embed a text collection; fit mini-batch k-means (batch size 32, k = number of ground-truth labels); score cluster assignments against labels with V-measure.

Models reporting this benchmark

These figures come from the model cards, which carry one collection date per card and no per-score attribution. They are shown as reported, not as verified evidence.
ModelProviderScoreCard as of
NV Embed v2NVIDIA53.82026-04
bge m3BAAI52.32026-04
Qwen3 Embedding 8BAlibaba / Qwen Team52.02026-04
Qwen3 VL Embedding 8BAlibaba / Qwen Team52.02026-04
jina embeddings v4Jina AI51.52026-04
jina embeddings v4 vllm retrievalJina AI51.52026-04
Voyage 3Voyage AI51.52026-04
e5 mistral 7B instructintfloat50.82026-04
bge multilingual gemma2BAAI50.52026-04
Gemini Embedding 001Google DeepMind50.52026-04
jina embeddings v3Jina AI50.22026-04
snowflake arctic embed l v2.0Snowflake49.82026-04
nomic embed text v2 moeNomic AI49.52026-04
nomic embed text v2 moe GGUFNomic AI49.52026-04
Qwen3 Embedding 4BAlibaba / Qwen Team49.52026-04
text-embedding-3-largeOpenAI49.22026-04
SFR Embedding 2 RSalesforce49.02026-04
bge large zh v1.5BAAI48.52026-04
multilingual e5 large instructintfloat48.52026-04
bge large en v1.5BAAI48.22026-04
bge large zhBAAI47.52026-04
granite embedding english r2IBM47.52026-04
granite embedding small english r2IBM47.52026-04
snowflake arctic embed lSnowflake47.22026-04
bge large enBAAI47.02026-04
Voyage Code 3Voyage AI47.02026-04
nomic embed text v1.5Nomic AI46.82026-04
nomic embed text v1.5 GGUFNomic AI46.82026-04
Voyage Finance 2Voyage AI46.82026-04
e5 large v2intfloat46.52026-04
snowflake arctic embed m v2.0Snowflake46.52026-04
Voyage Multilingual 2Voyage AI46.52026-04
Voyage 3 LiteVoyage AI46.22026-04
Voyage Law 2Voyage AI46.02026-04
bge base en v1.5BAAI45.82026-04
bge base zh v1.5BAAI45.52026-04
granite embedding 278M multilingualIBM45.52026-04
nomic embed text v1Nomic AI45.52026-04
multilingual e5 largeintfloat45.22026-04
snowflake arctic embed m v1.5Snowflake45.22026-04
e5 largeintfloat45.02026-04
bge base zhBAAI44.82026-04
jina embeddings v2 base enJina AI44.82026-04
snowflake arctic embed m longSnowflake44.82026-04
bge base enBAAI44.52026-04
Mistral EmbedMistral AI44.52026-04
e5 large unsupervisedintfloat44.22026-04
jina embeddings v2 base deJina AI44.22026-04
snowflake arctic embed mSnowflake44.22026-04
nomic embed codeNomic AI44.02026-04
Qwen3 Embedding 0.6BAlibaba / Qwen Team44.02026-04
Qwen3 VL Embedding 2BAlibaba / Qwen Team44.02026-04
e5 base v2intfloat43.82026-04
bge code v1BAAI43.52026-04
jina embeddings v2 base codeJina AI43.52026-04
SFR Embedding Code 400M RSalesforce43.22026-04
multilingual e5 baseintfloat43.02026-04
e5 baseintfloat42.52026-04
text-embedding-3-smallOpenAI42.52026-04
jina embeddings v2 small enJina AI42.22026-04
bge small en v1.5BAAI42.12026-04
bge small zh v1.5BAAI422026-04
granite embedding 107M multilingualIBM42.02026-04
granite embedding 125M englishIBM42.02026-04
snowflake arctic embed sSnowflake41.82026-04
all mpnet base v2Sentence Transformers41.52026-04
e5 small v2intfloat41.52026-04
bge small zhBAAI41.22026-04
multilingual e5 smallintfloat41.02026-04
all MiniLM L12 v2Sentence Transformers40.52026-04
bge small enBAAI40.52026-04
all MiniLM L6 v2Sentence Transformers40.12026-04
e5 smallintfloat40.02026-04
multi qa mpnet base dot v1Sentence Transformers39.82026-04
text-embedding-ada-002OpenAI39.82026-04
all roberta large v1Sentence Transformers39.52026-04
multi qa mpnet base cos v1Sentence Transformers39.52026-04
snowflake arctic embed xsSnowflake39.52026-04
paraphrase mpnet base v2Sentence Transformers39.22026-04
all distilroberta v1Sentence Transformers39.02026-04
paraphrase multilingual mpnet base v2Sentence Transformers38.52026-04
multi qa MiniLM L6 cos v1Sentence Transformers38.02026-04
granite embedding 30M englishIBM37.52026-04
granite embedding 30M sparseIBM37.52026-04
paraphrase multilingual MiniLM L12 v2Sentence Transformers37.52026-04
paraphrase MiniLM L12 v2Sentence Transformers37.22026-04
paraphrase MiniLM L6 v2Sentence Transformers36.82026-04
msmarco bert base dot v5Sentence Transformers36.52026-04
LaBSESentence Transformers36.22026-04
msmarco MiniLM L12 cos v5Sentence Transformers36.02026-04
msmarco MiniLM L6 v3Sentence Transformers35.02026-04
paraphrase MiniLM L3 v2Sentence Transformers34.52026-04
distiluse base multilingual cased v2Sentence Transformers34.02026-04
distiluse base multilingual cased v1Sentence Transformers32.52026-04

Data

This page as JSON · Edit on GitHub