Compatibility

pgvector is supported by EDB on EDB Postgres Advanced Server (EPAS), EDB Postgres Extended Server (PGE), and community PostgreSQL. Supported versions, platform requirements, and differences from community PostgreSQL are detailed in the sections below.

Supported database versions

This table lists supported pgvector versions and their corresponding EDB Postgres Advanced Server (EPAS), EDB Postgres Extended Server (PGE), and PostgreSQL (PG) versions.

pgvectorEPAS, PGE and PG 18EPAS, PGE and PG 17EPAS, PGE and PG 16EPAS, PGE and PG 15EPAS, PGE and PG 14EPAS, PGE and PG 13
0.8.5YYYYYY
0.8.4YYYYYY
0.8.3YYYYYY
0.8.2YYYYYY
0.8.1YYYYYY
0.8.0NYYYYY

For the full list of supported operating systems and CPU architectures, see:

Compatibility with EDB Postgres Distributed (PGD)

pgvector is compatible with EDB Postgres Distributed (PGD). Vector data stored in vector columns replicates normally through PGD's logical replication. However, there are important operational differences to be aware of — see Operations for details on index management and backup in PGD environments.

Compatibility with WarehousePG

pgvector is also supported on EDB Postgres AI for WarehousePG, a petabyte-scale MPP database. Because WarehousePG uses a distributed architecture, pgvector behavior differs from single-node Postgres deployments — for example, indexes must be managed per segment. For installation, configuration, and usage guidance specific to WarehousePG, see pgvector in WarehousePG.

Differences from community PostgreSQL

EDB-packaged pgvector is functionally identical to the upstream open-source extension. The following table summarizes feature availability across distributions and any pgvector-specific behavior.

FeaturePostgreSQLPGEEPASNotes
Package installationCommunity packagesEDB repository (RPM/DEB)EDB repository (RPM/DEB)EDB packages are tested against each supported Postgres version
Enterprise supportEDB support covers pgvector when used with PGE or EPAS
Row-level security (RLS)Standard PostgreSQL RLS policies apply to tables with vector columns
SSL/TLSAll existing SSL/TLS configuration applies to vector queries without additional setup
Auditingvia pgauditvia pgauditvia edb_auditVector query activity is logged alongside all other SQL operations
Encryption at rest (TDE)Vector data in tables and indexes is encrypted at rest under TDE; no application changes required
Oracle-compatible SQLpgvector works alongside EPAS Oracle-compatible features without conflicts

For known limitations, see Known limitations.


Could this page be better? Report a problem or suggest an addition!