AIDB 6.0.0 release notes v7
Released: 10 Feb 2026
AIDB 6.0.0 includes the following enhancements and bug fixes:
| Type | Description |
|---|---|
| Enhancement | Generic pipeline autoprocessing: Pipelines can now be configured to process data automatically in the background as new rows are inserted into the source table, without requiring manual invocation. |
| Enhancement | halfvec support: Added native support for the halfvec type (16-bit floating point vectors), reducing storage requirements for vector data by up to 50% compared to vector. |
| Enhancement | VectorChord integration: Introduced VectorChord as an integrated vector index backend, providing an alternative high-performance index type for similarity search. |
| Enhancement | Merge pipelines: Multiple pipelines can now be merged, enabling combined processing workflows across different pipeline configurations. |
| Enhancement | Intermediate step tables: Pipeline intermediate step results are now stored in dedicated tables, improving debuggability and enabling inspection of data at each stage of a multistep pipeline. |
| Enhancement | Pipeline source/destination migration: Added tooling to migrate existing pipeline source and destination table configurations to updated schemas. |
| Enhancement | Improved pipeline error messages: Error messages for invalid pipeline source table configurations are now more descriptive and actionable. |
| Change | Standardized pipeline destination table schemas: Pipeline destination tables now follow a standardized schema. Existing pipelines may require migration. See the migration guide for details. |