Decaf Blog Data pipelines, open-source tooling, reverse engineering, and systems notes

dbt-core v1.11.12 - ModelState Schema Fix

3 min read

dbt-core v1.11.12 was published on July 1, 2026, with a focused fix for state model config validation. The project now includes the ModelState schema in its bundled jsonschemas, which stops false deprecation warnings for state model configs such as pre_clone, lag_tolerance, and require_fresh_data_from. For ELT teams that run dbt in CI, scheduled jobs, or platform wrappers, this release is mostly about warning quality and operational signal.

The full release notes and downloads are on the GitHub release page.

Bundled schema coverage for ModelState

The release notes name one fix, tracked as #13016. dbt-core now adds the ModelState schema to the bundled jsonschemas shipped with the project. That matters because warning behavior depends on the schema set that dbt uses to understand model config fields.

Before this fix, valid state model configs could be reported as deprecated. The notes call these warnings spurious, which is the right reading for operators: the issue was not that the config keys had suddenly become invalid. The bundled schema metadata did not include the shape needed for these state config fields, so warning output drifted from the actual model contract.

This is an internal validation metadata change, not a new config surface. The public behavior is simpler: configs that are valid for state should stop looking deprecated just because the schema bundle was incomplete.

Cleaner signal for state model configs

The named config keys are concrete: pre_clone, lag_tolerance, require_fresh_data_from, and related state configs. If you watch dbt output for warning budgets, that list is the practical scope of the release.

False deprecation warnings are not harmless in a production ELT estate. They teach teams to ignore warnings, they mask real deprecations, and they create work during version audits. A schema fix like this keeps deprecation output closer to the actual contract of the version you are running.

For platform operators, this also helps wrappers that parse dbt logs. Many teams route warnings into CI annotations, Slack alerts, or run metadata tables. When state configs are valid but still reported as deprecated, those systems record noise. v1.11.12 reduces that noise without changing model SQL or materialization behavior.

Upgrade notes

There are no breaking changes or migration steps mentioned in the release notes. The safe reading is that v1.11.12 is a patch release for teams already on the 1.11 line and using state model configs. It is not a prerelease. That makes the release most relevant to teams that gate deployments on clean warning output.

The change is still worth testing in CI if your deployment treats warnings as failures or uses log parsers. The expected result is narrower output, not different compiled SQL. A useful check is to run the same project before and after the upgrade and compare warning counts for the named state config keys.

Where to get it