Cohere Advanced Retrieval (Embed & Rerank)
What is Cohere Advanced Retrieval (Embed & Rerank)?
Cohere Advanced Retrieval combines the Cohere Embed and Rerank model families into a developer-facing retrieval layer. Embed converts text, images, and mixed-format content into vector representations; Rerank orders retrieved documents according to their semantic relevance to a query.
Key Capabilities
- Semantic Retrieval: Embed produces vector representations for similarity search across enterprise content.
- Multimodal Search: Current Embed models support text, images, and mixed-modality business documents.
- Result Refinement: Rerank scores and reorders candidate documents after an initial lexical or semantic search.
- RAG and Agent Context: Both model families can supply relevant organizational information to agents and generative applications.
Audience & Use Cases
- Audience: Search engineers, AI application developers, data platform teams, and organizations building RAG systems.
- Use Cases: Semantic search, document retrieval, multimodal discovery, relevance optimization, and agent grounding.
Technical Specifications
- Embed 4: Supports text and image inputs, a 128,000-token context window, and configurable embedding dimensions.0
- Rerank 4: Pro and Fast variants support multilingual text and semi-structured data, with different quality and latency profiles.
- Deployment: Cohere APIs and private enterprise deployment options.
Categories & Use Cases
Technical Details
| Mobile Application | No |
|---|
FAQs
What is Cohere Advanced Retrieval (Embed & Rerank)?
Cohere Advanced Retrieval combines the Cohere Embed and Rerank model families into a developer-facing retrieval layer. Embed converts text, images, and mixed-format content into vector representations; Rerank orders retrieved documents according to their semantic relevance to a query.