Provider pricing
Mixedbread AI API pricing
Mixedbread (mxbai) ships the mxbai-embed and mxbai-rerank model families — top-of-MTEB embeddings and rerankers with native Matryoshka representation learning, so you can truncate dimensions for 4× faster vector search with minimal quality loss. Available as open weights and a hosted API.
Founded2023
HeadquartersBerlin, Germany
Tracked modelsDirectory
Best for
RAG and semantic-search teams needing high-quality embeddings without OpenAI-tier costs.
Strengths
- Top MTEB scores on mxbai-embed-large-v1 and -large-v2
- Matryoshka embeddings — truncate to 256 or 512 dims
- mxbai-rerank-large for high-precision RAG reranking
- Open weights on Hugging Face plus managed API