Postgres 18 Beta is out - Here are 7 features you should know about

Changelog

The latest product updates from Neon

RSS feed

Neon Local for local Postgres development with Docker

Announcing Neon Local, a new proxy service that lets you spin up and tear down isolated, production-like Postgres branches right from your local machine or CI, using Docker.

  • Automatically creates a new branch when your container starts and cleans it up when you're done
  • Works with any Postgres client, including the Neon serverless driver
  • Handles routing and authentication to your cloud database without manual configuration

Instantly create and destroy ephemeral Postgres environments with Docker and Neon Local.

Read the docs to learn more.

Beta: Import Data Assistant now automates your database migration to Neon

We're also excited to anncounce beta support for our new automated Import Data Assistant — a faster, simpler way to move your existing Postgres database to Neon. Just provide your connection string, and the assistant will handle the import for you, creating a new branch with your data.

  • Supports Postgres databases up to 10GB
  • Checks compatibility and guides you through the process

Import Data Assistant wizard

The Import Data Assistant guides you through the process of moving your data to Neon.

This feature is in beta and has some limitations (see docs for details). We're working to expand support for larger databases and more providers. If you try it, please let us know how it works for you — your feedback will help shape the future of Neon's import experience.

Read the docs to learn more.

Fixes & improvements
  • Neon Console

    • Fixed an issue where the connection string for a read replica could sometimes display the main (read-write) replica's connection string in the Connect modal. The correct connection string is now always shown.

    • Fixed an issue where the Console could display a read replica as the primary compute (or vice versa) in the Computes list on the Branch details page. This made it unclear which instance you were managing or observing. The correct compute is now always shown under the correct label.

    • Moved the branch selector to the sidebar for easier access. This and other recent changes are part of laying the groundwork for a more streamlined Console navigation experience coming soon – stay tuned!

      Branch selector in the sidebar

  • pg_search

    Updated the pg_search extension to version 0.15.18. See pg_search for details.

Neon Snapshots now available in Early Access

We're very happy to announce Early Access to Neon Snapshots, a powerful new way to capture and restore point-in-time copies of your database. Snapshots let you preserve your database state before making changes, running migrations, or simply bookmark a stable state.

Backup branch on the Branches page

With this update, we've also revamped our Backup & Restore page to provide a unified experience for both snapshots and instant restore (PITR) operations. Join our Early Access Program to try it out.

Read more about Neon Snapshots in our blog and the docs.

Support for scheduled snapshots, custom retention periods, and API/CLI integration coming later this year.

Postgres logs support for Datadog

We've added beta support for Postgres log exports to Datadog. You can now stream and analyze your database logs directly in your Datadog dashboard for better, centralized observability. Available on Scale and Business plans.

For more information, see Datadog Integration with Neon.

pg_search extension now available on Postgres 14, 15, and 16

We've updated the pg_search extension to version 0.15.16 and expanded support to Postgres 14, 15, and 16 (previously only available on Postgres 17). This powerful full-text search extension enables you to implement advanced search capabilities in your applications with minimal configuration.

To use pg_search, you'll need to load the required libraries with a simple Update project API call before installing the extension. Detailed instructions are available in our documentation: Enable the pg_search extension.

For more information and guidance, check out our guides:

Fixes & improvements
  • Neon Console

    • You can now find the Monitoring page under the Branches list in the sidebar. This is a bit of polish to our navigation: you now access Monitoring within your selected branch. If you want to monitor a different branch, use the main breadcrumb selector to change branches.
    • Fixed an issue where you sometimes could not access a branch if the branch creator's account was deleted. You can now view and access these branches normally.
    • Added a warning message when editing compute settings to help you plan for potential connection interruptions and temporary performance impacts when changing compute size.
  • Neon MCP Server

    • Released version 0.3.7 with improved Neon Auth setup instructions and compatibility with the latest Serverless Driver (1.0.0).
  • Read replica compute limit on the Free plan

    • To ensure consistent performance, we've introduced a limit of 3 read replica computes per project on the Free plan. This change helps maintain stability while still supporting common read scaling and analytics use cases.
  • PgBouncer version update

    • The PgBouncer version used by Neon to offer pooled connection support was updated to version 1.24.1

New default project setup

New Neon projects now start with a better out-of-the-box setup to support your dev workflow.

Instead of a single main branch, you'll now get:

  • A production branch (the default), designed for your production workload. It’s configured with a larger compute size (1–4 CU).
  • A development branch, created as a child of production, intended for local development. It uses a smaller compute size (0.25–1 CU).

new project production and development branches

This new project default aligns with typical usage scenarios, where your production branch will need more compute power (vCPU and RAM) than your less active development branches—but if you need something different, you can change your branch setup or compute sizes at any time.

To learn more about integrating branching into your dev workflow, read our Database branching workflow primer.

Neon MCP Server on Zed

You can now use the Neon MCP Server on Zed, a next-generation AI-powered code editor. For setup instructions, see Get started with Zed and Neon Postgres MCP Server.

MCP support in Zed is currently in preview. You can download the preview version of Zed from zed.dev/releases/preview.

Fixes & improvements
  • Neon MCP Server

    • The Neon MCP Server previously defaulted to the neondb_owner role when no Postgres role is provided, resulting in database access failures. It now uses the owner of the selected database instead. If a non-existent role is specified, the tool fails as expected.

      If no database name is provided, the server first looks for the Neon-created neondb database; if not found, it falls back to the first available database.

  • Neon Console

    • Updated plan descriptions on the Billing page to include root branch limits for each plan.
    • Added support for enabling HIPAA for existing Neon projects. Previously, HIPAA support could only be enabled for newly created Neon projects. Neon offers HIPAA compliance as part of our Business and Enterprise plans. For details, see HIPAA Compliance.
    • Added a warning to the Edit compute drawer in the Neon Console to inform users that changing compute size settings may briefly interrupt database connections.
    • The default AWS region for new projects created in the Neon Console is now AWS US East 1 (N. Virginia), instead of AWS US East 2 (Ohio).
  • Neon API

    • Added a new Create auth user API that lets users of Neon Auth add new users to the neon_auth.users_sync table. Newly created users are automatically propagated to your auth project, whether Neon-managed or provider-owned.

    • Changed the default AWS region for new Neon projects created via the Create project API. If no region_id is specified, the default is now aws-us-east-1 (N. Virginia), instead of aws-us-east-2 (Ohio).

    • The logical_size_bytes quota in the Create project and Update project APIs sets a storage limit for each branch. Previously, exceeding this limit prevented the branch's compute from starting. Now, computes can still start even when the quota is exceeded—only write operations are blocked. This allows users to delete data and bring usage back under the limit.

      The change applies automatically when setting a new logical_size_bytes value via the Update project API, or on the next compute restart for projects with a pre-existing quota.

  • Fixes

    • Fixed an issue in the Neon Console where branches created by a deleted user account couldn't be accessed. Attempting to open the branch returned a "Request failed" error.
    • Resolved an issue on the Project Dashboard where RAM usage was incorrectly shown in GiB instead of GB.
    • Resolved an issue in the Neon Postgres Previews Integration on Vercel where branches with child branches were incorrectly marked as obsolete. The automatic branch detection logic now checks for child branches.
    • Fixed an issue in the Native Vercel integration where the wrong password was set in Vercel preview environment variables if the default branch was defined as a protected branch.
Was this page helpful?

Subscribe to our changelog.
No spam, guaranteed.