rankings / text-embeddings
The best embedding models
44 models & services · 1 callable here now
Ranked by benchmark score per dollar (quality floor applied). Scores: BEIR (MTEB) — retrieval quality across the BEIR suite (NDCG@10 x 100). Fetched 2026-07-12. Prices are our live per-call rates; ~ marks an estimate until the model is onboarded.
sorted by value · sort by score
| # | model | score | price | params | status |
|---|---|---|---|---|---|
| 1 | intfloat/multilingual-e5-large-instruct best value | 52.7 (#41) | $0.0167 / 1M tokens | 560M | live · use now |
| 2 | voyageai/voyage-3-m-exp top score | 68.1 (#1) | - | 6918M | not hosted |
| 3 | TencentBAC/Conan-embedding-v2 | 66.4 (#2) | - | - | not hosted |
| 4 | nvidia/NV-Embed-v2 | 63.2 (#3) | - | 7850M | not hosted |
| 5 | Qwen/Qwen3-Embedding-8B | 62.8 (#4) | - | 7567M | not hosted |
| 6 | BAAI/bge-en-icl | 62.2 (#5) | - | 7110M | not hosted |
| 7 | yibinlei/LENS-d8000 | 61.9 (#6) | - | 7110M | not hosted |
| 8 | infly/inf-retriever-v1 | 61.7 (#7) | - | 7069M | not hosted |
| 9 | Qwen/Qwen3-Embedding-4B | 61.6 (#8) | - | 4022M | not hosted |
| 10 | yibinlei/LENS-d4000 | 60.8 (#9) | - | 7110M | not hosted |
| 11 | Linq-AI-Research/Linq-Embed-Mistral | 60.2 (#10) | - | 7111M | not hosted |
| 12 | Salesforce/SFR-Embedding-2_R | 59.8 (#11) | - | 7111M | not hosted |
| 13 | zeta-alpha-ai/Zeta-Alpha-E5-Mistral | 59.5 (#12) | - | 7111M | not hosted |
| 14 | NovaSearch/stella_en_1.5B_v5 | 59.3 (#13) | - | 1540M | not hosted |
| 15 | BAAI/bge-multilingual-gemma2 | 59.2 (#14) | - | 9240M | not hosted |
| 16 | Salesforce/SFR-Embedding-Mistral | 59.1 (#15) | - | 7111M | not hosted |
| 17 | Alibaba-NLP/gte-Qwen2-7B-instruct | 58.9 (#16) | - | 7069M | not hosted |
| 18 | BeastyZ/e5-R-mistral-7b | 58.7 (#17) | - | 7242M | not hosted |
| 19 | openbmb/MiniCPM-Embedding | 58.6 (#18) | - | 2725M | not hosted |
| 20 | infly/inf-retriever-v1-1.5b | 58.4 (#19) | - | 1543M | not hosted |
| 21 | Alibaba-NLP/gte-Qwen2-1.5B-instruct | 58.3 (#20) | - | 1543M | not hosted |
| 22 | codefuse-ai/F2LLM-v2-14B | 58.3 (#21) | - | 13990M | not hosted |
| 23 | NovaSearch/stella_en_400M_v5 | 57.9 (#22) | - | 435M | not hosted |
| 24 | codefuse-ai/F2LLM-v2-8B | 57.5 (#23) | - | 7568M | not hosted |
| 25 | lightonai/LateOn | 57.2 (#24) | - | 149M | not hosted |
| 26 | intfloat/e5-mistral-7b-instruct | 57.1 (#25) | - | 7111M | not hosted |
| 27 | codefuse-ai/F2LLM-v2-4B | 56.9 (#26) | - | 4022M | not hosted |
| 28 | McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised | 56.6 (#27) | - | 7547M | not hosted |
| 29 | Alibaba-NLP/gte-Qwen1.5-7B-instruct | 56.2 (#28) | - | 7099M | not hosted |
| 30 | lightonai/DenseOn | 56.2 (#29) | - | 149M | not hosted |
| 31 | McGill-NLP/LLM2Vec-Mistral-7B-Instruct-v2-mntp-supervised | 56 (#30) | - | 7153M | not hosted |
| 32 | Snowflake/snowflake-arctic-embed-l | 56 (#31) | - | 335M | not hosted |
| 33 | google/text-embedding-005 | 55.8 (#32) | - | - | not hosted |
| 34 | Alibaba-NLP/gme-Qwen2-VL-7B-Instruct | 55.7 (#33) | - | 7746M | not hosted |
| 35 | Qwen/Qwen3-Embedding-0.6B | 55.5 (#34) | - | 596M | not hosted |
| 36 | Snowflake/snowflake-arctic-embed-m-v2.0 | 55.5 (#35) | - | 305M | not hosted |
| 37 | google/text-embedding-004 | 55.5 (#36) | - | - | not hosted |
| 38 | openai/text-embedding-3-large | 55.4 (#37) | - | - | not hosted |
| 39 | lightonai/ColBERT-Zero | 55.4 (#38) | - | 149M | not hosted |
| 40 | Snowflake/snowflake-arctic-embed-l-v2.0 | 55.2 (#39) | - | 568M | not hosted |
| 41 | Alibaba-NLP/gte-modernbert-base | 55.2 (#40) | - | 149M | not hosted |
| 42 | OpenAI text-embedding-3-small | - | $0.02 / M tokens | - | reference |
| 43 | Cohere embed-v4 | - | $0.12 / M tokens | - | reference |
| 44 | Voyage voyage-3.5 | - | $0.06 / M tokens | - | reference |