Sep 24, 2026: Code Bundles (Preview)
Snowflake Code Bundles adds two execution options in preview: running Python code directly on warehouse compute and submitting Scala, Java, or Python Spark applications as batch jobs. Spark execution is built on Snowpark Connect. Scheduling notebooks on compute pools, previously offered as Notebook Projects, is now generally available. Inline specification overrides and the REST API, Python API, and Snowflake CLI clients are also in preview.
Read more
What changed
Code Bundles adds warehouse execution for Python with compute_type: warehouse and batch execution for Spark applications with type: spark, both in preview. Notebook scheduling on compute pools with Snowpark Container Services is now generally available. Inline specification overrides and REST API, Python API, and Snowflake CLI clients remain preview capabilities.
Why it matters
Code Bundles can execute non-SQL code such as Python without wrapping it in a stored procedure. Warehouse execution and Spark jobs expand the options for running code on Snowflake.
Practical impact
The update is relevant to engineers running Python workloads or existing Spark applications on Snowflake. Warehouse and Spark execution are in preview; notebook scheduling is generally available, so these availability stages should be distinguished when considering adoption.