Mixedbread

Introducing Mixedbread Wholembed v3, our new SOTA retrieval model across all modalities and 100+ languages. Wholembed v3 brings best-in-class search to text, audio, images, PDFs, videos... You can now get the best retrieval performance on your data, no matter its format.

Build your own granular citation with Mixedbread:

Our high-quality processing detects the layout of documents, further supporting: - granular QA citations - your downstream LLM understanding

New: Granular evidence citations in QA Beyond chunk-level citations, our QA now points to the exact evidence snippet supporting each answer. Easier to inspect, verify, and trust. → Available with high-quality processing.

@mixedbreadai's wholembed-v3 set a high bar for video retrieval as a single-stage retriever on our MAGMaR shared task! To beat it, both C2F-RAG and MARQUIS needed multi-stage pipelines with reasoning-based reranking on top. That's a strong model.

Powerful search in your control:

New: Bring your own bucket Mixedbread is built on top of object storage. Now that storage can be yours. Your data stays in a bucket you control. Mixedbread indexes and searches with zero content retention on our side. For enterprise teams that need control, compliance, audit visibility, and simpler security reviews.

Agents can inspect file metadata in a store to understand available filters. docs: mixedbread.com/docs/stores/se…

New: Metadata explorer Adding metadata to files enables filtering during search. Now, you can browse metadata fields and values across your store.

Read more here: mixedbread.com/blog/latent-te…

Having language-adjacent properties means that tools designed for lexical approaches "just work". BM25, always refusing to exit the scene, is strong here: applied over the Latent Terms extracted from nomic-embed-v1.5, it results in a near state-of-the-art sparse retriever.

By now, everyone knows that single-vector embedding models are hugely limiting for modern workflows. But they contain than you think: you can extract sparse Latent Terms from them. And it turns out that BM25 is all you need to turn this vocabulary into a strong retriever.

docs: mixedbread.com/api-reference/…

New: grep for exact matching grep → keyword / regex matching search → fine-grained semantic retrieval Works across uploaded content, including text, PDFs (OCR) and audio/video (transcription). Give your agents both retrieval primitives to perform at their best.

View and export traces directly from your dashboard:

Steer search with more instructions. Docs: mixedbread.com/docs/stores/se…

Feature: Native agentic search on Mixedbread Search with auto-planning, exploration, and multi-hop reasoning across documents. Built for: - evidence discovery - exhaustive search - cross-document reasoning → Topped MADQA @snowflake with 93.4% accuracy across 18,000 PDF pages.

New: Traces for Mixedbread agentic search See every search call an agent makes directly in the dashboard, and tune instructions for better retrieval quality.

Read more here: mixedbread.com/blog/listwise-…

Introducing mxbai-rerank-v3-listwise: reranking that goes beyond binary relevance. It reads the whole candidate set, resolves conflicts, and ranks by directives like recency, source priority, and multi-step rules. +11% NDCG@10 on average across multiple domains, modalities, and languages in runs with Wholembed v3. Available today in preview in Mixedbread.