We just released an API that makes it easy to add versioning and checkpoints to your agent platform
/Changelog

Changelog

The latest product updates from Neon

RSS feed

Snapshot APIs for database versioning (Beta)

We've added snapshot APIs for AI agents and code generation platforms that need database versioning. You can create point-in-time database versions, roll back when things break, and keep connection strings stable during development. This makes it safer for agents to experiment with schema changes and revert when needed. Works for production rollbacks as well as temporary preview branches.

# Create a snapshot to save current state
curl --request POST \
     --url 'https://console.neon.tech/api/v2/projects/{project_id}/branches/{branch_id}/snapshot?name=version-session-1&expires_at=2025-08-13T00:00:00Z' \
     --header 'authorization: Bearer $NEON_API_KEY'

# Roll back to a previous version (keeps same connection string)
curl --request POST \
     --url 'https://console.neon.tech/api/v2/projects/{project_id}/snapshots/{snapshot_id}/restore' \
     --header 'authorization: Bearer $NEON_API_KEY' \
     --header 'content-type: application/json' \
     --data '{"target_branch_id": "br-main-123", "finalize_restore": true}'

Learn more in these guides:

Building a full-stack AI agent? Apply to our Agents Program for custom limits and special pricing.

Data API setup gets easier with automated grants

Our latest addition to the beta Data API makes setup even smoother. The Data API configuration now includes an option to automatically apply GRANT statements for the public schema. When you enable "Grant public schema access to authenticated users", Neon handles the database permissions for you, setting you up to then write your RLS policies. See Get started with the Data API.

Data API grants configuration

Free plan compute hours → 100

We just doubled the compute hours you get in your Neon Free Plan. You get up to 20 projects, each with:

  • 100 CU-hours of monthly compute (up from 50)
  • Snapshot APIs
  • 0.5 GB Storage
  • Autoscaling
  • Branching
  • Read replicas
  • Instant restore

If you are a free plan user, this change has already been applied to your account — no action required.

Postgres extensions analytics dashboard

We've built an interactive dashboard to explore Postgres extension adoption across the Neon platform. See which extensions are trending, track monthly install counts, and discover new extensions that might be useful for your projects. The dashboard shows real-time data on 83+ available extensions, from popular ones like pgvector (30k+ installs) to emerging extensions gaining traction.

Postgres Extensions Analytics dashboard

Self-serve HIPAA compliance

You can now enable HIPAA compliance for your Neon projects directly in the console. HIPAA sets national standards for protecting health information, and is required for apps handling Protected Health Information (PHI). This is available on the Scale plan. HIPAA support is currently available at no additional cost. Once billing is finalized, it will become a paid add-on. Read more in our HIPAA compliance docs.

Enable HIPAA

Neon Private Networking Adds IPv6 Support

Neon Private Networking, which allows secure connections via AWS PrivateLink and keeps traffic within AWS's private network, now includes support for IPv6 addresses. This enables better connectivity for applications that require IPv6 support and helps future-proof your database connections. Previously, only IPv4 was supported.

ChatGPT + Neon MCP Server

You can now connect ChatGPT to the Neon MCP Server using custom Model Context Protocol (MCP) connectors.

ChatGPT with Neon MCP Server

This integration makes it easy to extend ChatGPT with Neon’s database capabilities—so you can query, manage, and interact with your Neon projects directly within ChatGPT.

👉 Read the blog to get started.

Community: Neon Testing – Vitest library for integration tests

We’re excited to share Neon Testing, a community-built library by Mikael Lirbank that makes it easy to run integration tests against Neon databases.

Instead of relying on mocks or maintaining local test databases, Neon Testing uses Neon’s branching to provision disposable Postgres test databases for each test file. This means your tests run against the same schema and constraints as production, catching issues that mocks can miss (like unique constraint failures and transaction rollbacks).

Neon OTel integration + New Relic

Neon’s OpenTelemetry (OTEL) integration has a new step-by-step guide for sending your Neon project’s metrics and Postgres logs to New Relic—a cloud-based observability platform that helps developers and organizations monitor, debug, and optimize the performance of their applications and infrastructure.

Check out the guide from contributor Dhanush Reddy: Getting started with Neon and New Relic.

Improved Neon Docs navigation

We've introduced a new horizontal navigation bar in the Neon Docs to make it easier to find what you need. You can now quickly scan across the top-level menus, each with a dropdown of related topics. When you select a menu item, a dedicated left-hand sidebar for that topic area opens, giving you a clear view of everything available within that section.

This update improves discoverability and helps you move through the docs more efficiently.

Neon docs new navigation

Fixes & improvements
  • Neon Console
    • We adjusted the warning message on the Edit compute modal about connection disruptions when changing the compute size. The warning message now only appears when compute size values are modified.
    • Fixed an issue where the Branch expiration modal would close without notice if an error occurred. The modal now remains open and displays the error message.
  • Backup & restore
    • On the Backup & restore page in the Neon Console, snapshots are now listed with a more user-friendly branch name instead of the branch ID value.
    • The Restore branch modal now shows the new branch expiration time that will be set when restoring a branch configured to expire.
  • Native Vercel integration
    • On the Branch overview page for users of the native Vercel integration, the Open preview deployment link now correctly directs to Vercel deployment page instead of the application page.
    • You can now open the Branch overview page in the Neon Console for a preview deployment branch directly from the Vercel deployment page.

Introducing the Neon Agent Plan

We're happy to announce our new Agent Plan for AI agent platforms that need to provision thousands of databases.

agent plan from console

The Agent Plan provides custom project and branch limits, higher API rate limits for instant provisioning, and credits for your free tier users. It's designed for platforms like Replit, v0, and Databutton that create databases on behalf of their users. Key features include automatic scale-to-zero, instant restores, and Neon Auth and Data API at no extra cost. Learn more.

Data API — more improvements

We've added a Refresh schema cache button to the Data API page, making it easier to update your REST endpoints when you modify your database schema. We've also continued our performance enhancements, including eliminating API cold starts, as we prepare the Data API for GA. Learn more about the Data API in the docs.

data api refresh schema cache button

Neon MCP Server now with reset_from_parent tool

We've added a new reset_from_parent tool to the Neon MCP Server that allows resetting a branch back to its parent branch state. This simplifies branch management when LLMs change schemas or when creating fresh development branches. Learn more in the MCP Server docs.

New VPC endpoint service for Private Networking

We've added a new VPC endpoint service address for Private Networking in the AWS Europe (Frankfurt) region. If you're setting up Private Networking in the eu-central-1 region, you can now use the additional endpoint service address for enhanced infrastructure capacity and reliability.

For the complete list of VPC endpoint service addresses by region, see our Private Networking guide.

Fixes & improvements
  • Neon Console
    • Added restart compute functionality to compute management menus, letting you restart compute instances directly from the Console without using CLI or API

      new compute restart button

    • Added a Compute last active at column to your organization's Projects table, helping you identify unused projects for cleanup and cost optimization

      compute last active column

    • Fixed password encoding in Parameters only connection strings, ensuring special characters display correctly for environment variable usage.

      parameters only connection string

    • Added URL encoding for projects list page search and pagination, making it easier to share specific views with team members. For example:

      console.neon.tech/app/org-example/projects?cursor=example-branch-123&q=myapp
Was this page helpful?

Subscribe to our changelog.
No spam, guaranteed.