Mixedbread

Mixedbread builds open embedding and reranking models that power retrieval for search and retrieval augmented generation systems. Its models rank highly on public embedding benchmarks and are released with open weights alongside a hosted API for teams that prefer a managed service. The company works on multimodal and multilingual retrieval, which is the part of the stack that determines whether a language model has the right context to answer; its research team publishes evaluation results and training details with each release.