dbt Core v1.12 Now Generally Available with Rust Parser and New Workflow Features

dbt Labs has announced the general availability of dbt Core v1.12, a release that pairs practical improvements for current users with an early, low-risk path toward the upcoming v2.0 major version.

The headline feature is an opt-in Rust-based parser, activated with the --use-v2-parser flag, which replaces the existing Python parser for project compilation and DAG construction. On larger projects, this new parser can be 5-10 times faster. Because the flag is entirely optional and can be removed at any time, teams can test it against their real projects and address any compatibility issues gradually. This parser is the same one that will power dbt Core v2.0, making v1.12 a deliberate stepping stone to that release.

Beyond the parser, v1.12 introduces several workflow enhancements. A new on_error config gives teams control over whether downstream models are skipped (the default) or continue running after an upstream failure, addressing a community request dating back to 2020. Project variables can now live in a dedicated vars.yml file at the project root, keeping dbt_project.yml cleaner and making those variables available during project file parsing. A new --sql flag for dbt run-operation allows ad hoc SQL execution through dbt's Jinja context without creating a named macro.

User-defined functions (UDFs) also see notable expansion. JavaScript UDFs are now supported for Snowflake and BigQuery, Python UDFs can run on Databricks (with Unity Catalog), the new overloads property allows a single function to accept multiple argument signatures, and Python UDFs can declare public PyPI packages for automatic installation.

The release also includes an updated Semantic Layer specification that nests semantic definitions directly within models, plus support for Apache Ossie documents (formerly Open Semantic Interchange) as an alternative format. Adapter-specific improvements cover Snowflake (Iceberg v3, dynamic-table controls), BigQuery (parallel microbatch execution), Redshift (query_groupsession parameter), and Databricks (row filters, tag merging).

Other quick hits include native private package installation via existing SSH config, a simplified catalogs.yml spec, composable selectors, clearer error messages, and an optional pointer view for versioned models. The dbt team will host a live virtual recap and Q&A to walk through the release and discuss the roadmap to v2.0.

Source: dbt Labs Blog