Mixedbread

Mixedbread is a search and retrieval platform aimed at developers building AI agents and LLM applications that need to find information across mixed content types. It handles the full pipeline — uploading, parsing, embedding, and indexing — for text, PDFs, images, video, and audio in more than 100 languages, so teams do not have to manage embedding models or vector infrastructure themselves. The company built its reputation on open-source embedding and reranking models; its flagship mxbai-embed-large-v1 model has passed 5.5 million downloads on Hugging Face and scores ahead of OpenAI's text-embedding-3-large on the MTEB benchmark. On top of these models, the hosted platform offers semantic search, keyword/regex search, question answering, and metadata filtering through a REST API, Python and TypeScript SDKs, a CLI for bulk syncing, and a Model Context Protocol integration that connects directly to tools like Claude Code. Mixedbread reports sub-200ms search latency and throughput above 10,000 queries per second in production, and holds SOC2 Type II and ISO 27001 certifications, with on-premise and regional deployment options for compliance-sensitive customers. Named customers include Vercel, Deepset, and AnkiHub. Pricing starts with a free tier including one-time credits, moving to usage-based paid plans; the company does not disclose a founding year on its site.