dbt

dbt is how analytics teams turn raw warehouse tables into trustworthy, documented data models, using plain SQL plus the engineering discipline analytics always lacked: version control, testing, and dependency tracking. You write SELECT statements, and dbt works out the run order, tests the output, and builds the docs. The open source core is Apache 2.0.

2026 rewired the project. dbt 2.0 went GA in September as a ground-up Rust rewrite that ships as a single binary with no Python runtime. The naming changed with it: the Apache 2.0 subset is now dbt OSS, and the full dbt binary is proprietary but free to use, including the VS Code extension, whose old 15-user cap is gone. dbt 1.x gets one final minor release, 1.13, with critical fixes promised for several years.

Solo analysts and small teams: dbt OSS plus your own scheduler (Airflow, Dagster, cron) is the free, complete path. The hosted dbt platform is free for one developer, then Starter at $100 per user per month for up to five seats. Enterprise, with Mesh, SSO, and SCIM, is custom pricing.

The catch is who owns it. dbt Labs merged into Fivetran in June 2026. The combined company says it will maintain dbt OSS under Apache 2.0 indefinitely, but the best developer experience now lives in a proprietary binary whose terms bar reverse engineering. Build on the open subset if that matters to you.