MTEB SciFact
28 models with a score. Leader: Qwen3 Embedding 8B with 78.5.
| Benchmark profile | |
|---|---|
| Task | Retrieval (en) |
| Metric | nDCG@10 |
| Direction | ↑ higher is better |
| Limitation | Ein Wert gilt fuer die gemessene Modellrevision und die volle Vektorlaenge; gekuerzte Laengen sind nicht gemessen (R-MESS). |
| Source | huggingface.co |
| Licence | CC0 — MTEB-Ergebnisse (embeddings-benchmark/results) |
Ranking (28)
No entry matches the filter.
| Model | Rank | Score | Developer | EU processing: InputEU in | EU processing: OutputEU out | all providers: Inputall in | all providers: Outputall out | Open weights | GPU memory (4-bit, derived)GPU memory |
|---|---|---|---|---|---|---|---|---|---|
| Qwen3 Embedding 8B | 1 | 78.5 | Alibaba Cloud | €0.009 | – | €0.009 | – | ✓ yes | from 9 GB |
| E5-Mistral-7B | 2 | 76.4 | Microsoft | €0.02 | – | €0.02 | – | ✓ yes | from 8 GB |
| bge-large-en-v1.5 | 3 | 74.6 | BAAI | – | – | €0.009 | – | ✓ yes | from 5 GB |
| bge-base-en-v1.5 | 4 | 74.3 | BAAI | – | – | €0.0045 | – | ✓ yes | from 5 GB |
| mxbai-embed-large-v1 | 5 | 73.9 | Mixedbread | – | – | – | – | ✓ yes | from 5 GB |
| snowflake-arctic-embed-l | 6 | 73.8 | Snowflake | – | – | – | – | ✓ yes | from 5 GB |
| gte-multilingual-base | 7 | 73.4 | Alibaba Cloud | – | – | – | – | ✓ yes | from 5 GB |
| bekko-embedding-v1-a25m | 8 | 72.3 | hotchpotch | – | – | – | – | ✓ yes | from 5 GB |
| snowflake-arctic-embed-m-v2.0 | 8 | 72.3 | Snowflake | – | – | €0.036 | – | ✓ yes | from 5 GB |
| BGE Multilingual Gemma2 | 10 | 72 | BAAI | €0.01 | – | €0.01 | – | ✓ yes | from 10 GB |
| SNOWFLAKE ARCTIC-EMBED-M-V1.5 | 11 | 71.84 | Snowflake | – | – | €0.022 | – | ✓ yes | from 5 GB |
| multilingual-e5-large-instruct | 12 | 71.6 | Microsoft | €0.03 | €0.02 | €0.009 | – | ✓ yes | from 5 GB |
| snowflake-arctic-embed-l-v2.0 | 13 | 70.9 | Snowflake | – | – | – | – | ✓ yes | from 5 GB |
| granite-embedding-311m-multilingual-r2 | 14 | 70.8 | IBM | – | – | – | – | ✓ yes | from 5 GB |
| Amazon Titan Text Embeddings V2 | 15 | 70.5 | Amazon | – | – | – | – | – no | – |
| nomic-embed-text-v1.5 | 16 | 70.3 | Nomic AI | – | – | – | – | ✓ yes | from 5 GB |
| intfloat/multilingual-e5-large | 17 | 70.2 | Microsoft | €0.0082 | – | €0.0082 | – | ✓ yes | from 5 GB |
| bekko-embedding-v1-a8m | 18 | 70 | hotchpotch | – | – | – | – | ✓ yes | from 5 GB |
| cohere-embed-multilingual-v3.0 | 18 | 70 | Cohere | – | – | €0.09 | – | – no | – |
| snowflake-arctic-embed-m-long | 18 | 70 | Snowflake | – | – | – | – | ✓ yes | from 5 GB |
| Qwen3-Embedding-0.6B | 21 | 69.7 | Alibaba Cloud | – | – | €0.011 | – | ✓ yes | from 5 GB |
| granite-embedding-97m-multilingual-r2 | 22 | 68.7 | IBM | – | – | – | – | ✓ yes | from 5 GB |
| multilingual-e5-small | 23 | 67.58 | – | – | €0.0067 | – | ✓ yes | from 5 GB | |
| granite-embedding-278m-multilingual | 24 | 65.8 | IBM | – | – | – | – | ✓ yes | from 5 GB |
| all-mpnet-base-v2 | 25 | 65.57 | – | – | – | €0.0045 | – | ✓ yes | from 5 GB |
| all-MiniLM-L6-v2 | 26 | 64.5 | – | – | – | €0.0045 | – | ✓ yes | from 5 GB |
| BGE-M3 | 27 | 64.4 | BAAI | €0.0082 | – | €0.0082 | – | ✓ yes | from 5 GB |
| All MiniLM L12 v2 | 28 | 62.64 | – | – | – | €0.0045 | – | ✓ yes | from 5 GB |
Equal scores share a rank. When filtering, the rank stays the position in the whole list. GPU memory “from …” is a lower bound (parameters × 0.55 GB + 4 GB). Prices: € per 1M tokens. Source: providers' pricing pages, read and checked by searchit. As of 7 Nov 2026.
Scores: MTEB SciFact, CC0 — MTEB-Ergebnisse (embeddings-benchmark/results). Taken unchanged from the source.