Llama Embed Nemotron 8B vs Qwen 3 Embedding 4B
MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages) Embeddings assign examples to categories, group similar examples, estimate semantic similarity between pairs, classify relationships between pairs, rank relevant documents for queries, find translated sentence pairs, rerank documents according to instructions, assign multiple labels to each example, reorder candidate documents by relevance on MTEB Multilingual v2 · mean over tasks in 1037 languages including 'Auhelawa, Abau, Achi, Achinese, Achuar-Shiwiar, Adzera, Afrikaans, Agarabi; tasks receive equal weight
Llama Embed Nemotron 8B69.5
Qwen 3 Embedding 4B69.4
FollowIR · Mean over tasks · English Embeddings rerank documents according to instructions on FollowIR · mean over tasks in English; tasks receive equal weight
Llama Embed Nemotron 8B10.8
Qwen 3 Embedding 4B11.6
FollowIR · Mean over task types · English Embeddings rerank documents according to instructions on FollowIR · mean over task types in English; task types receive equal weight
Llama Embed Nemotron 8B10.8
Qwen 3 Embedding 4B11.6
FollowIR · Instruction Reranking · English Embeddings rerank documents according to instructions on FollowIR · Instruction Reranking in English
Llama Embed Nemotron 8B10.8
Qwen 3 Embedding 4B11.6
MTEB Europe v1 · Bitext Mining · multilingual (28 languages) Embeddings find translated sentence pairs on MTEB Europe v1 · Bitext Mining in 28 languages including Basque, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian
Llama Embed Nemotron 8B92.1
Qwen 3 Embedding 4B92.2
MTEB Europe v1 · Clustering · multilingual (29 languages) Embeddings group similar examples on MTEB Europe v1 · Clustering in 29 languages including Basque, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian
Llama Embed Nemotron 8B49.9
Qwen 3 Embedding 4B50.4
MTEB Europe v1 · Instruction Reranking · English Embeddings rerank documents according to instructions on MTEB Europe v1 · Instruction Reranking in English
Llama Embed Nemotron 8B10.8
Qwen 3 Embedding 4B11.6
MTEB Europe v1 · Multilabel Classification · multilingual (23 languages) Embeddings assign multiple labels to each example on MTEB Europe v1 · Multilabel Classification in 23 languages including Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish
Llama Embed Nemotron 8B22
Qwen 3 Embedding 4B23.6
MTEB Europe v1 · Reranking · multilingual (12 languages) Embeddings reorder candidate documents by relevance on MTEB Europe v1 · Reranking in Bulgarian, Czech, Danish, Dutch, English, Finnish, French, German, Italian, Portuguese, Romanian, Swedish
Llama Embed Nemotron 8B62.5
Qwen 3 Embedding 4B62
MTEB French v1 · Pair Classification · French Embeddings classify relationships between pairs on MTEB French v1 · Pair Classification in French
Llama Embed Nemotron 8B66.8
Qwen 3 Embedding 4B72.6
MTEB Indic v1 · Clustering · multilingual (21 languages) Embeddings group similar examples on MTEB Indic v1 · Clustering in Assamese, Awadhi, Bengali, Bhojpuri, Chhattisgarhi, Gujarati, Hindi, Kannada, Kashmiri, Maithili, Malayalam, Manipuri, Marathi, Nepali (individual language), Odia, Panjabi, Sanskrit, Sindhi, Tamil, Telugu, Urdu
Llama Embed Nemotron 8B57
Qwen 3 Embedding 4B49.7
MTEB Indic v1 · Pair Classification · Hindi Embeddings classify relationships between pairs on MTEB Indic v1 · Pair Classification in Hindi
Llama Embed Nemotron 8B80.1
Qwen 3 Embedding 4B85.6
MTEB Indic v1 · Reranking · multilingual (2 languages) Embeddings reorder candidate documents by relevance on MTEB Indic v1 · Reranking in Bengali, Hindi
Llama Embed Nemotron 8B88.5
Qwen 3 Embedding 4B85.8
MTEB Indic v1 · STS · multilingual (13 languages) Embeddings estimate semantic similarity between pairs on MTEB Indic v1 · STS in Assamese, Bengali, English, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Panjabi, Tamil, Telugu, Urdu
Llama Embed Nemotron 8B58.2
Qwen 3 Embedding 4B60.8
MTEB Korean v2 · Retrieval · multilingual (2 languages) Embeddings rank relevant documents for queries on MTEB Korean v2 · Retrieval in English, Korean
Llama Embed Nemotron 8B56.3
Qwen 3 Embedding 4B77.3
MTEB multilingual v2 · Mean over task types · multilingual (1037 languages) Embeddings assign examples to categories, group similar examples, estimate semantic similarity between pairs, classify relationships between pairs, rank relevant documents for queries, find translated sentence pairs, rerank documents according to instructions, assign multiple labels to each example, reorder candidate documents by relevance on MTEB Multilingual v2 · mean over task types in 1037 languages including 'Auhelawa, Abau, Achi, Achinese, Achuar-Shiwiar, Adzera, Afrikaans, Agarabi; task types receive equal weight
Llama Embed Nemotron 8B61.1
Qwen 3 Embedding 4B60.9
MTEB multilingual v2 · Bitext Mining · multilingual (1026 languages) Embeddings find translated sentence pairs on MTEB Multilingual v2 · Bitext Mining in 1026 languages including 'Auhelawa, Abau, Achi, Achinese, Achuar-Shiwiar, Adzera, Afrikaans, Agarabi
Llama Embed Nemotron 8B81.7
Qwen 3 Embedding 4B79.4
MTEB multilingual v2 · Classification · multilingual (119 languages) Embeddings assign examples to categories on MTEB Multilingual v2 · Classification in 119 languages including Achinese, Afrikaans, Albanian, Algerian Arabic, Amharic, Arabic, Armenian, Assamese
Llama Embed Nemotron 8B73.2
Qwen 3 Embedding 4B72.3
MTEB multilingual v2 · Clustering · multilingual (208 languages) Embeddings group similar examples on MTEB Multilingual v2 · Clustering in 208 languages including Achinese, Afrikaans, Akan, Albanian, Amharic, Armenian, Assamese, Asturian
Llama Embed Nemotron 8B54.4
Qwen 3 Embedding 4B57.1
MTEB multilingual v2 · Instruction Reranking · English Embeddings rerank documents according to instructions on MTEB Multilingual v2 · Instruction Reranking in English
Llama Embed Nemotron 8B10.8
Qwen 3 Embedding 4B11.6
MTEB multilingual v2 · Multilabel Classification · multilingual (25 languages) Embeddings assign multiple labels to each example on MTEB Multilingual v2 · Multilabel Classification in 25 languages including Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish
Llama Embed Nemotron 8B29.9
Qwen 3 Embedding 4B26.8
MTEB multilingual v2 · Pair Classification · multilingual (24 languages) Embeddings classify relationships between pairs on MTEB Multilingual v2 · Pair Classification in 24 languages including Arabic, Armenian, Bulgarian, Chinese, Czech, English, Finnish, French
Llama Embed Nemotron 8B84
Qwen 3 Embedding 4B85.1
MTEB multilingual v2 · Reranking · multilingual (22 languages) Embeddings reorder candidate documents by relevance on MTEB Multilingual v2 · Reranking in 22 languages including Bengali, Bulgarian, Czech, Danish, Dutch, English, Finnish, French
Llama Embed Nemotron 8B67.8
Qwen 3 Embedding 4B65.1
MTEB multilingual v2 · Retrieval · multilingual (120 languages) Embeddings rank relevant documents for queries on MTEB Multilingual v2 · Retrieval in 120 languages including Afrikaans, Amharic, Arabic, Armenian, Assamese, Bambara, Basque, Bengali
Llama Embed Nemotron 8B68.7
Qwen 3 Embedding 4B69.6
MTEB multilingual v2 · STS · multilingual (35 languages) Embeddings estimate semantic similarity between pairs on MTEB Multilingual v2 · STS in 35 languages including Afrikaans, Algerian Arabic, Amharic, Arabic, Assamese, Bengali, Dutch, English
Llama Embed Nemotron 8B79.4
Qwen 3 Embedding 4B80.9
MTEB Dutch v1 · Mean over tasks · Dutch Embeddings assign examples to categories, group similar examples, rank relevant documents for queries, estimate semantic similarity between pairs, assign multiple labels to each example, classify relationships between pairs, reorder candidate documents by relevance on MTEB Dutch v1 · mean over tasks in Dutch; tasks receive equal weight
Llama Embed Nemotron 8B66.2
Qwen 3 Embedding 4B63.7
MTEB Dutch v1 · Mean over task types · Dutch Embeddings assign examples to categories, group similar examples, rank relevant documents for queries, estimate semantic similarity between pairs, assign multiple labels to each example, classify relationships between pairs, reorder candidate documents by relevance on MTEB Dutch v1 · mean over task types in Dutch; task types receive equal weight
Llama Embed Nemotron 8B70.7
Qwen 3 Embedding 4B68.5
MTEB Dutch v1 · Classification · Dutch Embeddings assign examples to categories on MTEB Dutch v1 · Classification in Dutch
Llama Embed Nemotron 8B73.4
Qwen 3 Embedding 4B70.7
MTEB Dutch v1 · Clustering · Dutch Embeddings group similar examples on MTEB Dutch v1 · Clustering in Dutch
Llama Embed Nemotron 8B50.8
Qwen 3 Embedding 4B46.5
MTEB Dutch v1 · Multilabel Classification · Dutch Embeddings assign multiple labels to each example on MTEB Dutch v1 · Multilabel Classification in Dutch
Llama Embed Nemotron 8B41.5
Qwen 3 Embedding 4B40.2
MTEB Dutch v1 · Pair Classification · Dutch Embeddings classify relationships between pairs on MTEB Dutch v1 · Pair Classification in Dutch
Llama Embed Nemotron 8B83.9
Qwen 3 Embedding 4B81.5
MTEB Dutch v1 · Reranking · Dutch Embeddings reorder candidate documents by relevance on MTEB Dutch v1 · Reranking in Dutch
Llama Embed Nemotron 8B91.7
Qwen 3 Embedding 4B89.1
MTEB Dutch v1 · Retrieval · Dutch Embeddings rank relevant documents for queries on MTEB Dutch v1 · Retrieval in Dutch
Llama Embed Nemotron 8B67
Qwen 3 Embedding 4B65.1
MTEB Dutch v1 · STS · Dutch Embeddings estimate semantic similarity between pairs on MTEB Dutch v1 · STS in Dutch
Llama Embed Nemotron 8B86.4
Qwen 3 Embedding 4B86.1
MTEB Portuguese v1 · Multilabel Classification · Portuguese Embeddings assign multiple labels to each example on MTEB Portuguese v1 · Multilabel Classification in Portuguese
Llama Embed Nemotron 8B29
Qwen 3 Embedding 4B20.6
MTEB Russian v1.1 · Pair Classification · Russian Embeddings classify relationships between pairs on MTEB Russian v1.1 · Pair Classification in Russian
Llama Embed Nemotron 8B67.2
Qwen 3 Embedding 4B66.6
MTEB Scandinavian v1 · Bitext Mining · multilingual (3 languages) Embeddings find translated sentence pairs on MTEB Scandinavian v1 · Bitext Mining in Danish, Norwegian Bokmål, Norwegian Nynorsk
Llama Embed Nemotron 8B79.6
Qwen 3 Embedding 4B82.2
MTEB Spanish v1 · Pair Classification · Spanish Embeddings classify relationships between pairs on MTEB Spanish v1 · Pair Classification in Spanish
Llama Embed Nemotron 8B74.8
Qwen 3 Embedding 4B80.1
MTEB Thai v1 · Clustering · Thai Embeddings group similar examples on MTEB Thai v1 · Clustering in Thai
Llama Embed Nemotron 8B63.4
Qwen 3 Embedding 4B54.3
MTEB Thai v1 · Pair Classification · Thai Embeddings classify relationships between pairs on MTEB Thai v1 · Pair Classification in Thai
Llama Embed Nemotron 8B81.4
Qwen 3 Embedding 4B86.7
BEIR · Mean over tasks · English Embeddings rank relevant documents for queries on BEIR · mean over tasks in English; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B61.6
BEIR · Mean over task types · English Embeddings rank relevant documents for queries on BEIR · mean over task types in English; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B61.6
BEIR · Retrieval · English Embeddings rank relevant documents for queries on BEIR · Retrieval in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B61.6
CoIR · Mean over tasks · multilingual (9 languages) Embeddings rank relevant documents for queries on CoIR · mean over tasks in English, c++, go, java, javascript, php, python, ruby, sql; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B79.2
CoIR · Mean over task types · multilingual (9 languages) Embeddings rank relevant documents for queries on CoIR · mean over task types in English, c++, go, java, javascript, php, python, ruby, sql; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B79.2
CoIR · Retrieval · multilingual (9 languages) Embeddings rank relevant documents for queries on CoIR · Retrieval in English, c++, go, java, javascript, php, python, ruby, sql
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B79.2
CoREB v1 · Mean over tasks · multilingual (6 languages) Embeddings reorder candidate documents by relevance, rank relevant documents for queries on CoReB v1 · mean over tasks in English, c++, go, java, python, ruby; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B56.5
CoREB v1 · Mean over task types · multilingual (6 languages) Embeddings reorder candidate documents by relevance, rank relevant documents for queries on CoReB v1 · mean over task types in English, c++, go, java, python, ruby; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B56.5
CoREB v1 · Reranking · multilingual (6 languages) Embeddings reorder candidate documents by relevance on CoReB v1 · Reranking in English, c++, go, java, python, ruby
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B52.3
CoREB v1 · Retrieval · multilingual (6 languages) Embeddings rank relevant documents for queries on CoReB v1 · Retrieval in English, c++, go, java, python, ruby
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B60.6
MTEB Mandarin Chinese v1 · Mean over tasks · Mandarin Chinese Embeddings rank relevant documents for queries, estimate semantic similarity between pairs, reorder candidate documents by relevance, group similar examples, classify relationships between pairs, assign examples to categories on MTEB Chinese v1 · mean over tasks in Mandarin Chinese; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B72.3
MTEB Mandarin Chinese v1 · Mean over task types · Mandarin Chinese Embeddings rank relevant documents for queries, estimate semantic similarity between pairs, reorder candidate documents by relevance, group similar examples, classify relationships between pairs, assign examples to categories on MTEB Chinese v1 · mean over task types in Mandarin Chinese; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B73.5
MTEB Mandarin Chinese v1 · Classification · Mandarin Chinese Embeddings assign examples to categories on MTEB Chinese v1 · Classification in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B75.5
MTEB Mandarin Chinese v1 · Clustering · Mandarin Chinese Embeddings group similar examples on MTEB Chinese v1 · Clustering in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B77.9
MTEB Mandarin Chinese v1 · Pair Classification · Mandarin Chinese Embeddings classify relationships between pairs on MTEB Chinese v1 · Pair Classification in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B83.3
MTEB Mandarin Chinese v1 · Reranking · Mandarin Chinese Embeddings reorder candidate documents by relevance on MTEB Chinese v1 · Reranking in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B66.1
MTEB Mandarin Chinese v1 · Retrieval · Mandarin Chinese Embeddings rank relevant documents for queries on MTEB Chinese v1 · Retrieval in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B77
MTEB Mandarin Chinese v1 · STS · Mandarin Chinese Embeddings estimate semantic similarity between pairs on MTEB Chinese v1 · STS in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B61.3
MTEB Code v1 · Mean over tasks · multilingual (15 languages) Embeddings rank relevant documents for queries on MTEB Code v1 · mean over tasks in English, c, c++, go, java, javascript, php, python, ruby, rust, scala, shell, sql, swift, typescript; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B80.1
MTEB Code v1 · Mean over task types · multilingual (15 languages) Embeddings rank relevant documents for queries on MTEB Code v1 · mean over task types in English, c, c++, go, java, javascript, php, python, ruby, rust, scala, shell, sql, swift, typescript; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B80.1
MTEB Code v1 · Retrieval · multilingual (15 languages) Embeddings rank relevant documents for queries on MTEB Code v1 · Retrieval in English, c, c++, go, java, javascript, php, python, ruby, rust, scala, shell, sql, swift, typescript
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B80.1
MTEB Korean v2 · Mean over tasks · multilingual (2 languages) Embeddings estimate semantic similarity between pairs, rank relevant documents for queries, classify relationships between pairs, reorder candidate documents by relevance, assign examples to categories, group similar examples on MTEB Korean v2 · mean over tasks in English, Korean; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B76.9
MTEB Korean v2 · Mean over task types · multilingual (2 languages) Embeddings estimate semantic similarity between pairs, rank relevant documents for queries, classify relationships between pairs, reorder candidate documents by relevance, assign examples to categories, group similar examples on MTEB Korean v2 · mean over task types in English, Korean; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B74.7
MTEB Korean v2 · Classification · Korean Embeddings assign examples to categories on MTEB Korean v2 · Classification in Korean
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B67.4
MTEB Korean v2 · Clustering · Korean Embeddings group similar examples on MTEB Korean v2 · Clustering in Korean
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B70.9
MTEB Korean v2 · Pair Classification · Korean Embeddings classify relationships between pairs on MTEB Korean v2 · Pair Classification in Korean
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B75.3
MTEB Korean v2 · Reranking · Korean Embeddings reorder candidate documents by relevance on MTEB Korean v2 · Reranking in Korean
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B67.3
MTEB Korean v2 · STS · Korean Embeddings estimate semantic similarity between pairs on MTEB Korean v2 · STS in Korean
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B89.9
MTEB Medical v1 · Clustering · English Embeddings group similar examples on MTEB Medical v1 · Clustering in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B44.6
MTEB Medical v1 · Reranking · Mandarin Chinese Embeddings reorder candidate documents by relevance on MTEB Medical v1 · Reranking in Mandarin Chinese
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B85.1
MTEB Russian v1.1 · Mean over tasks · Russian Embeddings assign examples to categories, estimate semantic similarity between pairs, reorder candidate documents by relevance, group similar examples, classify relationships between pairs, rank relevant documents for queries, assign multiple labels to each example on MTEB Russian v1.1 · mean over tasks in Russian; tasks receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B70.1
MTEB Russian v1.1 · Mean over task types · Russian Embeddings assign examples to categories, estimate semantic similarity between pairs, reorder candidate documents by relevance, group similar examples, classify relationships between pairs, rank relevant documents for queries, assign multiple labels to each example on MTEB Russian v1.1 · mean over task types in Russian; task types receive equal weight
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B67.5
MTEB Russian v1.1 · Classification · Russian Embeddings assign examples to categories on MTEB Russian v1.1 · Classification in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B73.6
MTEB Russian v1.1 · Clustering · Russian Embeddings group similar examples on MTEB Russian v1.1 · Clustering in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B65.1
MTEB Russian v1.1 · Multilabel Classification · Russian Embeddings assign multiple labels to each example on MTEB Russian v1.1 · Multilabel Classification in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B43.3
MTEB Russian v1.1 · Reranking · Russian Embeddings reorder candidate documents by relevance on MTEB Russian v1.1 · Reranking in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B68.7
MTEB Russian v1.1 · Retrieval · Russian Embeddings rank relevant documents for queries on MTEB Russian v1.1 · Retrieval in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B76.4
MTEB Russian v1.1 · STS · Russian Embeddings estimate semantic similarity between pairs on MTEB Russian v1.1 · STS in Russian
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B78.5
MTEB Scandinavian v1 · Classification · multilingual (6 languages) Embeddings assign examples to categories on MTEB Scandinavian v1 · Classification in Danish, Faroese, Icelandic, Norwegian Bokmål, Norwegian Nynorsk, Swedish
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B72.9
MTEB Scandinavian v1 · Retrieval · multilingual (3 languages) Embeddings rank relevant documents for queries on MTEB Scandinavian v1 · Retrieval in Danish, Norwegian Bokmål, Swedish
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B68.4
MTEB Thai v1 · Classification · Thai Embeddings assign examples to categories on MTEB Thai v1 · Classification in Thai
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B72.4
MTEB Thai v1 · Reranking · Thai Embeddings reorder candidate documents by relevance on MTEB Thai v1 · Reranking in Thai
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B61.7
MTEB English v2 · Pair Classification · English MTEB English v2 · Pair Classification · English: embeddings classify relationships between pairs in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B87
MTEB English v2 · Classification · English MTEB English v2 · Classification · English: embeddings assign examples to categories in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B89.8
MTEB English v2 · Reranking · English MTEB English v2 · Reranking · English: embeddings reorder candidate documents by relevance in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B50.8
MTEB English v2 · STS · English MTEB English v2 · STS · English: embeddings estimate semantic similarity between pairs in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B88.7
MTEB English v2 · Retrieval · English MTEB English v2 · Retrieval · English: embeddings rank relevant documents for queries in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B68.5
MTEB English v2 · Clustering · English MTEB English v2 · Clustering · English: embeddings group similar examples in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B57.5
MTEB English v2 · Summarization · English MTEB English v2 · Summarization · English: embeddings match summaries to their source meaning in English
Llama Embed Nemotron 8B—
Qwen 3 Embedding 4B34.4
| Benchmark | Llama Embed Nemotron 8B | Qwen 3 Embedding 4B |
|---|---|---|
| MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages) | 69.5 | 69.4 |
| FollowIR · Mean over tasks · English | 10.8 | 11.6 |
| FollowIR · Mean over task types · English | 10.8 | 11.6 |
| FollowIR · Instruction Reranking · English | 10.8 | 11.6 |
| MTEB Europe v1 · Bitext Mining · multilingual (28 languages) | 92.1 | 92.2 |
| MTEB Europe v1 · Clustering · multilingual (29 languages) | 49.9 | 50.4 |
| MTEB Europe v1 · Instruction Reranking · English | 10.8 | 11.6 |
| MTEB Europe v1 · Multilabel Classification · multilingual (23 languages) | 22 | 23.6 |
| MTEB Europe v1 · Reranking · multilingual (12 languages) | 62.5 | 62 |
| MTEB French v1 · Pair Classification · French | 66.8 | 72.6 |
| MTEB Indic v1 · Clustering · multilingual (21 languages) | 57 | 49.7 |
| MTEB Indic v1 · Pair Classification · Hindi | 80.1 | 85.6 |
| MTEB Indic v1 · Reranking · multilingual (2 languages) | 88.5 | 85.8 |
| MTEB Indic v1 · STS · multilingual (13 languages) | 58.2 | 60.8 |
| MTEB Korean v2 · Retrieval · multilingual (2 languages) | 56.3 | 77.3 |
| MTEB multilingual v2 · Mean over task types · multilingual (1037 languages) | 61.1 | 60.9 |
| MTEB multilingual v2 · Bitext Mining · multilingual (1026 languages) | 81.7 | 79.4 |
| MTEB multilingual v2 · Classification · multilingual (119 languages) | 73.2 | 72.3 |
| MTEB multilingual v2 · Clustering · multilingual (208 languages) | 54.4 | 57.1 |
| MTEB multilingual v2 · Instruction Reranking · English | 10.8 | 11.6 |
| MTEB multilingual v2 · Multilabel Classification · multilingual (25 languages) | 29.9 | 26.8 |
| MTEB multilingual v2 · Pair Classification · multilingual (24 languages) | 84 | 85.1 |
| MTEB multilingual v2 · Reranking · multilingual (22 languages) | 67.8 | 65.1 |
| MTEB multilingual v2 · Retrieval · multilingual (120 languages) | 68.7 | 69.6 |
| MTEB multilingual v2 · STS · multilingual (35 languages) | 79.4 | 80.9 |
| MTEB Dutch v1 · Mean over tasks · Dutch | 66.2 | 63.7 |
| MTEB Dutch v1 · Mean over task types · Dutch | 70.7 | 68.5 |
| MTEB Dutch v1 · Classification · Dutch | 73.4 | 70.7 |
| MTEB Dutch v1 · Clustering · Dutch | 50.8 | 46.5 |
| MTEB Dutch v1 · Multilabel Classification · Dutch | 41.5 | 40.2 |
| MTEB Dutch v1 · Pair Classification · Dutch | 83.9 | 81.5 |
| MTEB Dutch v1 · Reranking · Dutch | 91.7 | 89.1 |
| MTEB Dutch v1 · Retrieval · Dutch | 67 | 65.1 |
| MTEB Dutch v1 · STS · Dutch | 86.4 | 86.1 |
| MTEB Portuguese v1 · Multilabel Classification · Portuguese | 29 | 20.6 |
| MTEB Russian v1.1 · Pair Classification · Russian | 67.2 | 66.6 |
| MTEB Scandinavian v1 · Bitext Mining · multilingual (3 languages) | 79.6 | 82.2 |
| MTEB Spanish v1 · Pair Classification · Spanish | 74.8 | 80.1 |
| MTEB Thai v1 · Clustering · Thai | 63.4 | 54.3 |
| MTEB Thai v1 · Pair Classification · Thai | 81.4 | 86.7 |
| BEIR · Mean over tasks · English | — | 61.6 |
| BEIR · Mean over task types · English | — | 61.6 |
| BEIR · Retrieval · English | — | 61.6 |
| CoIR · Mean over tasks · multilingual (9 languages) | — | 79.2 |
| CoIR · Mean over task types · multilingual (9 languages) | — | 79.2 |
| CoIR · Retrieval · multilingual (9 languages) | — | 79.2 |
| CoREB v1 · Mean over tasks · multilingual (6 languages) | — | 56.5 |
| CoREB v1 · Mean over task types · multilingual (6 languages) | — | 56.5 |
| CoREB v1 · Reranking · multilingual (6 languages) | — | 52.3 |
| CoREB v1 · Retrieval · multilingual (6 languages) | — | 60.6 |
| MTEB Mandarin Chinese v1 · Mean over tasks · Mandarin Chinese | — | 72.3 |
| MTEB Mandarin Chinese v1 · Mean over task types · Mandarin Chinese | — | 73.5 |
| MTEB Mandarin Chinese v1 · Classification · Mandarin Chinese | — | 75.5 |
| MTEB Mandarin Chinese v1 · Clustering · Mandarin Chinese | — | 77.9 |
| MTEB Mandarin Chinese v1 · Pair Classification · Mandarin Chinese | — | 83.3 |
| MTEB Mandarin Chinese v1 · Reranking · Mandarin Chinese | — | 66.1 |
| MTEB Mandarin Chinese v1 · Retrieval · Mandarin Chinese | — | 77 |
| MTEB Mandarin Chinese v1 · STS · Mandarin Chinese | — | 61.3 |
| MTEB Code v1 · Mean over tasks · multilingual (15 languages) | — | 80.1 |
| MTEB Code v1 · Mean over task types · multilingual (15 languages) | — | 80.1 |
| MTEB Code v1 · Retrieval · multilingual (15 languages) | — | 80.1 |
| MTEB Korean v2 · Mean over tasks · multilingual (2 languages) | — | 76.9 |
| MTEB Korean v2 · Mean over task types · multilingual (2 languages) | — | 74.7 |
| MTEB Korean v2 · Classification · Korean | — | 67.4 |
| MTEB Korean v2 · Clustering · Korean | — | 70.9 |
| MTEB Korean v2 · Pair Classification · Korean | — | 75.3 |
| MTEB Korean v2 · Reranking · Korean | — | 67.3 |
| MTEB Korean v2 · STS · Korean | — | 89.9 |
| MTEB Medical v1 · Clustering · English | — | 44.6 |
| MTEB Medical v1 · Reranking · Mandarin Chinese | — | 85.1 |
| MTEB Russian v1.1 · Mean over tasks · Russian | — | 70.1 |
| MTEB Russian v1.1 · Mean over task types · Russian | — | 67.5 |
| MTEB Russian v1.1 · Classification · Russian | — | 73.6 |
| MTEB Russian v1.1 · Clustering · Russian | — | 65.1 |
| MTEB Russian v1.1 · Multilabel Classification · Russian | — | 43.3 |
| MTEB Russian v1.1 · Reranking · Russian | — | 68.7 |
| MTEB Russian v1.1 · Retrieval · Russian | — | 76.4 |
| MTEB Russian v1.1 · STS · Russian | — | 78.5 |
| MTEB Scandinavian v1 · Classification · multilingual (6 languages) | — | 72.9 |
| MTEB Scandinavian v1 · Retrieval · multilingual (3 languages) | — | 68.4 |
| MTEB Thai v1 · Classification · Thai | — | 72.4 |
| MTEB Thai v1 · Reranking · Thai | — | 61.7 |
| MTEB English v2 · Pair Classification · English | — | 87 |
| MTEB English v2 · Classification · English | — | 89.8 |
| MTEB English v2 · Reranking · English | — | 50.8 |
| MTEB English v2 · STS · English | — | 88.7 |
| MTEB English v2 · Retrieval · English | — | 68.5 |
| MTEB English v2 · Clustering · English | — | 57.5 |
| MTEB English v2 · Summarization · English | — | 34.4 |
Specs
| Spec | Llama Embed Nemotron 8B | Qwen 3 Embedding 4B |
|---|---|---|
| Price | — | !$0.02 |
| Context | 33K | 33K |
| Weights | Open | Open |
| Released | Oct 2025 | Jun 2025 |
| Dimensions | 4,096 | 2,560 |
| Max input | 131K tokens | 33K tokens |
| Memory (4-bit) | 4.5 GB | 2.4 GB |
| Licence | https://huggingface.co/nvidia/llama-embed-nemotron-8b/blob/main/LICENSE | Apache-2.0 |
| Commercial use | Restricted | Allowed |
What breaks when you switch
Llama Embed Nemotron 8B → Qwen 3 Embedding 4B
Re-embed your whole corpus›
Vectors from different models can't be mixed; dimensions change from 4,096 to 2,560.
Max input: 131K tokens → 33K tokens
Qwen 3 Embedding 4B → Llama Embed Nemotron 8B
Re-embed your whole corpus›
Vectors from different models can't be mixed; dimensions change from 2,560 to 4,096.