Cortex Analyst

What it does

Cortex Analyst takes a business question written in plain language, along with a semantic model that describes your Snowflake tables, metrics and relationships. The standalone REST API generates SQL from the question and semantic model; the SQL is then executed in your Snowflake virtual warehouse to produce the final output, which the calling application handles. The semantic model can be a Semantic View (recommended) or a legacy YAML file stored on a stage.

People typically use it to build self-serve analytics chat experiences, where non-technical users ask questions without writing SQL. It is offered as a REST API so it can be called from Streamlit apps, Slack, Teams or custom interfaces. Snowflake recommends moving new work to Cortex Agents, which supports every Cortex Analyst capability; the standalone Analyst API remains documented.

This suits data and application teams already on Snowflake who need governed, role-based access to structured data. Callers need a role with the SNOWFLAKE.CORTEX_USER or SNOWFLAKE.CORTEX_ANALYST_USER database role plus SELECT on the tables referenced in the semantic model. Limitations include reliance on a well-built semantic model for accuracy, and that executing the generated SQL incurs virtual warehouse compute charges.

What it can do

• Answers natural-language business questions over structured Snowflake data
• Generates SQL from a natural-language question and semantic model; the generated SQL is executed in a Snowflake virtual warehouse
• Uses Semantic Views to define metrics, dimensions, relationships and examples
• Enforces Snowflake role-based access controls on generated queries
• Supports multi-turn follow-up questions
• Automatically selects among multiple data sources when configured

Before you choose

• Accuracy depends heavily on the quality of the semantic model: descriptions, metrics and verified examples.
• Pricing has two paths: Cortex Agents token-based AI Credits, or standalone Analyst API billed per 1,000 messages, plus warehouse compute for executed SQL.
• Snowflake recommends transitioning to Cortex Agents, which it states supports every Cortex Analyst capability.

Listed under

• Data
• Ask your data

Get started

Requires a role with Cortex Analyst database role access.

1. Create or identify a Semantic View (or legacy YAML semantic model) describing your data.
2. Grant the SNOWFLAKE.CORTEX_ANALYST_USER (or CORTEX_USER) database role to a custom or existing analyst role and assign it to users; optionally use CORTEX_ANALYST_USER for selective least-privilege access.
3. Grant SELECT on the referenced tables, plus READ on any stage holding a YAML model and USAGE on any Cortex Search services it mentions.
4. Authenticate API requests with an authorization token and send a test question to confirm a SQL response is returned.

Billed as usage, not a subscription. Via Cortex Agents, billed in AI Credits per million tokens plus warehouse compute for executed SQL. Via the standalone Analyst API, billed per 1,000 messages plus warehouse compute. Exact rates are in Snowflake's consumption table and pricing page; no flat per-seat fee applies.

Source: docs.snowflake.com/en/user-guide/snowflake-cortex/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Cortex Analyst | Snowflake Documentation ↗docs.snowflake.com
• Snowflake AI pricing | Snowflake Documentation ↗docs.snowflake.com

Company website

docs.snowflake.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

manual_review

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.