Snowflake Cortex Search

Overview

Snowflake Cortex Search is a managed service that provides a hybrid search approach to retrieving and ranking documents. It utilizes vector search for semantic similarity, keyword search for lexical matching, and semantic reranking for optimal relevance.

Recent Updates (March 2026)

• Support for pre-computed vector embeddings from third-party models
• Dynamic filters for applying filter conditions on each search call
• Dynamic columns to specify which metadata columns to retrieve per call
• Dynamic result count up to 500 results per call
• Enhanced hybrid retrieval capabilities

How It Works

Cortex Search takes a hybrid approach:

• Vector Search: Retrieves semantically similar documents using embeddings
• Keyword Search: Retrieves lexically similar documents using traditional search
• Semantic Reranking: Reranks the most relevant documents in the result set

Key Features

• Serverless architecture integrated with Snowflake Data Cloud
• Support for both Snowflake-provided embeddings and custom pre-computed vectors
• Dynamic search behavior control via API parameters
• Attribute-based filtering during search
• Metadata column selection per query
• Enterprise-grade security and governance

Cost Structure

• Embedding tokens charged during index creation
• Ongoing serving costs per GB of index maintained
• Different pricing model than AISQL functions
• Usage-based billing integrated with Snowflake credits

Use Cases

• Enterprise knowledge bases and document retrieval
• Semantic search over structured and unstructured data
• RAG applications with Snowflake data
• Compliance-sensitive AI applications requiring data governance

Integration

Fully integrated with Snowflake ecosystem including Snowpark, Streamlit, and Snowflake's security features.