Models
We build open-source embedding and reranking models for production use, with over 50 million downloads on Hugging Face. Our models consistently perform well on industry benchmarks while being designed for real-world applications like semantic search, RAG, and document retrieval.
Featured models
Check out some of our most popular and powerful models for embedding and reranking tasks.
mxbai-embed-large-v1
Our flagship embedding model
mxbai-rerank-large-v2
Our flagship reranking model
mxbai-rerank-xsmall-v1
Fast and affordable reranking
Embedding models
Our embedding models transform text into rich vector representations, powering semantic search, RAG, clustering, and more. Explore the different models available, including multilingual options and our Colbert model.
mxbai-embed-large-v1
Our flagship embedding model
deepset-mxbai-embed-de-large-v1
Collaborative DE/EN model
mxbai-colbert-large-v1
Our late interaction embedding model
mxbai-embed-2d-large-v1
Our 2D Matryoshka embedding model
mxbai-embed-xsmall-v1
Small but powerful embedding model
Reranking models
Enhance the precision of your search results with our reranking models. These models re-order initial candidate lists based on deep semantic relevance. Use them to boost the quality of your search and recommendation systems significantly.
mxbai-rerank-large-v2
Our flagship reranking model
mxbai-rerank-base-v2
Perfectly balanced for most use cases
mxbai-rerank-large-v1
Large rerank model from the first series
mxbai-rerank-base-v1
Base rerank model from the first series
mxbai-rerank-xsmall-v1
Fast and efficient reranking model
Last updated: June 17, 2025