Snowflake Cortex is the AI and machine learning capability layer inside the Snowflake Data Cloud, offering LLM functions, semantic search and fully managed models on your governed data.
Cortex brings generative and classic ML to where enterprise data already lives. Analysts can call LLM functions — complete, summarize, translate, extract, classify — directly in SQL, while engineers build retrieval-augmented applications with Cortex Search and Cortex Analyst over governed Snowflake tables. Because everything runs inside Snowflake’s security perimeter, data does not leave the platform.
Cortex also offers managed models for forecasting and anomaly detection, and hosts open models like Llama and Mistral for teams that want choice. Fine-tuning, semantic views and Snowflake Copilot round out a suite designed to make AI accessible to SQL-literate teams.
Cortex’s scope covers the full generative spectrum: text completion, summarization, translation, classification and extraction functions callable in SQL, plus Cortex Search for retrieval-augmented generation and Cortex Analyst for conversational access to structured data. Classic ML needs are served by forecasting and anomaly detection functions, and hosted open models give teams choice without infrastructure work. All of it runs under Snowflake’s governance model — role-based access, data masking and audit logging apply to AI usage like any other query.
Cortex functions are consumption-based within Snowflake credits, with separate pricing per feature; enterprise agreements are handled through Snowflake sales.
Organizations already on Snowflake that want LLM and ML capabilities on governed data without moving it to another AI platform.
Cortex is consumption-priced within Snowflake; some functions have free trial allowances during preview.
Yes — Cortex functions are callable from SQL and REST endpoints, and integrate with Snowpark for Python.
Its advantage is running AI where governed data already lives, eliminating the data-movement problem of external AI tools.
Yes — Cortex LLM and ML functions are callable directly in SQL, so AI processing can be embedded in pipelines, views and transformations.
Data teams use Cortex to summarize and classify text columns in SQL, build retrieval-augmented applications over governed data, provide natural language access to structured data through Cortex Analyst, and forecast metrics with managed models. All of it runs within Snowflake’s security model, keeping sensitive data in place while AI transforms it.
Metabase is an open-source business intelligence tool that lets anyone in a company ask questions of their data without SQL. OverviewMetabase makes…
Databricks is a unified data and AI platform combining a lakehouse architecture with analytics, machine learning, and generative AI tooling. OverviewDatabricks provides…
