Snowflake Cortex

Managed LLM and machine learning functions available inside Snowflake SQL

AI Data AnalysisOverseas★★★★☆ 4.0

Visit Snowflake Cortex official site ↗ View similar tools Compare alternatives →

What is Snowflake Cortex?

Snowflake Cortex is the AI layer inside the Snowflake data cloud. It exposes SQL functions for text generation, summarisation, classification, translation, sentiment analysis and embeddings, all running against data that never leaves the account. Cortex Analyst converts natural-language questions into SQL against a semantic model, Cortex Search provides hybrid retrieval for building assistants over documents, and the ML functions cover forecasting, anomaly detection and classification on tabular data. Because everything is invoked through SQL or a Python API, teams can add AI steps to existing pipelines without standing up separate infrastructure or managing model endpoints.

Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.

Key features

SQL functions for summarisation, classification, translation and sentiment
Embedding generation and vector search within the account
Cortex Analyst for natural-language questions against a semantic model
Cortex Search for hybrid retrieval over documents and text columns
ML functions for forecasting, anomaly detection and classification
Serverless inference with no endpoints or GPUs to manage

Pros & cons

Strengths

Data stays inside the account boundary during inference
Callable from ordinary SQL, so pipelines need no new infrastructure
No model serving or GPU capacity to provision and monitor

Watch out for

Model choice is limited to what Snowflake hosts and approves
Consumption-based credits make cost forecasting harder
Fine-tuning and custom architectures are outside its scope

Best for & use cases

warehouse-native ai, text analytics, document retrieval and sql forecasting

If you're comparing similar products, check the alternatives below, or browse all tools in the AI Data Analysis category.

FAQ

Do I need a separate AI platform to use this?

For the common tasks - summarising text, classifying rows, generating embeddings, forecasting a series - no. Everything is reachable through SQL. Custom training, fine-tuning or exotic model architectures still belong in a dedicated ML platform.

Is my data used for model training?

Snowflake positions these as inference functions over data within your account, and the models are hosted rather than trained on customer data. Confirm the contractual terms for your specific region and model before handling regulated data.

What does it cost in practice?

Inference consumes credits, and large tables can multiply that quickly. Sampling and trimming inputs, or materialising results once instead of recomputing per query, are the usual ways teams keep the bill controlled.