PostgreSQL 19 is targeting November 19, 2026, after the release team chose to include another round of fixes before general availability. PostgreSQL committer Robert Haas attributes the delay partly to AI-assisted bug discovery outpacing human review. The project's announcement confirms the schedule, while Haas's explanation is a contributor's assessment of the workload.
His account appeared at Neon on October 21 and was reposted on his blog on October 30. It gives developers planning an upgrade a reason to separate the release timetable from claims about how much work AI can automate.
More fixes still need human review
For April 8 through September 25, Haas reports 724 commits to REL_18_STABLE in 2025. For the same dates in 2026, he reports 1,169 commits to REL_19_STABLE. These are the respective release branches for those years. His comparison shows a larger maintenance workload, though commit totals alone cannot establish the number of distinct bugs or how many AI found.
Haas says model-generated patches still need people to check their correctness and compliance with PostgreSQL standards. Work on security issues and older branches also took time away from reviewing version 19. His explanation describes a review bottleneck, without measuring an AI tool's accuracy or the hours it saved.
What upgrade planners can use now
The October 15 release announcement schedules PostgreSQL 19 RC 2 for November 12, alongside updates for supported versions. The team says it wants those fixes included before version 19 ships. November 19 remains conditional: another issue could require a delay or an additional release candidate.
The same announcement says upgrading to RC 1 from an earlier version requires a major-version upgrade strategy, such as pg_upgrade or a dump and restore. For an upgrade rehearsal, that is a concrete requirement to account for before choosing a migration date.
Watch the November 12 release candidate and the project's release announcement for changes to the November 19 target. A confirmed general release will give deployment plans a firmer date than Haas's diagnosis of the review workload can supply.