Mixedbread

Legacy Embedding and Reranking Models

Mixedbread previously published standalone documentation for direct embedding and reranking APIs and for each open-source model generation. Those reference pages are no longer part of the current documentation, but the model artifacts and their technical release posts remain available.

Embedding modelsLink to section

Model familyRelease postResources
mxbai-embed-large-v1Technical overviewHugging Face
mxbai-embed-2d-large-v1Technical overviewHugging Face
mxbai-embed-xsmall-v1Technical overviewHugging Face
deepset-mxbai-embed-de-large-v1Technical overviewHugging Face
mxbai-colbert-large-v1Technical overviewHugging Face

Reranking modelsLink to section

Model familyRelease postResources
mxbai-rerank-*-v1Technical overviewHugging Face
mxbai-rerank-*-v2Technical overviewHugging Face

Current search workflowLink to section

For new managed-search integrations, start with Stores. Stores handle parsing, indexing, retrieval, and reranking as one workflow. See Store search for retrieval options and reranking for current model configuration.

If you maintain an integration with the earlier direct embedding or reranking APIs and need migration help, contact support@mixedbread.com.