Streamlit in Snowflake

Streamlit in Snowflake (SiS) is Snowflake's native runtime for Streamlit apps. The app's code is stored as a Snowflake object and runs on Snowflake compute with a live session against the account's data, so there is no Docker image, deployment pipeline, or connection string to manage.

It differs from Snowpark Container Services in scope. SiS is narrow and Streamlit-specific: write the Python script in Snowsight or push it from Git, and it runs where the data already is, governed by the same roles and policies. SPCS is a general container platform that can also host a containerised Streamlit app when more control is needed. SiS apps can run on a virtual warehouse or on a container runtime backed by a compute pool.

Apps are shared with other users in the account by granting a role, with separate viewer and builder access, which makes SiS a quick way to turn an analysis into an internal tool or dashboard without involving an infrastructure team. Python packages come from Snowflake's Anaconda channel or, on the container runtime, from PyPI.

Cost is billed in Snowflake credits for the warehouse or compute pool that runs the app. A warehouse stays active while viewers have the app open and for a short time afterwards, so popular apps can keep compute running. Like the rest of this family, it only exists as an option inside an existing Snowflake account.