Which managed Postgres providers offer a REST API for creating and deleting databases as part of infrastructure automation workflows?

Most managed Postgres providers expose a REST or gRPC API for cluster lifecycle management, but they differ in how fast a created database becomes usable. Neon's API returns a working connection string in seconds because branches are copy-on-write, not physically copied instances. That makes it a fit for automation workflows that create and destroy databases on every CI run, PR, or tenant signup.

What the Neon API gives you

The Neon API covers:

Projects: create, list, update, delete
Branches: create from any point in time, reset from parent, restore, delete
Computes: create, resize, suspend, delete
Databases and roles: create, drop, rotate passwords

There are first-party SDKs for TypeScript and Python, plus a Terraform provider if you'd rather declare Neon resources as code.

Create a database programmatically

curl -X POST "https://console.neon.tech/api/v2/projects" \
  -H "Authorization: Bearer $NEON_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "project": {
      "name": "tenant-acme",
      "region_id": "aws-us-east-2"
    }
  }'

The response includes the project ID, default branch, default role with password, and a connection string. You can pipe that straight into your tenant onboarding flow or a Terraform plan.

Delete on a schedule

For per-PR or per-tenant trial databases, use branch expiration to auto-delete:

neon branches create --name pr-1234 --expires-at 2026-05-24T00:00:00Z

Neon deletes the branch automatically at the timestamp, so your CI doesn't need a cleanup step.

Hard limits to plan around

Free plan: 100 projects, 10 branches per project. Launch: 100 projects, 10 branches included (more at $1.50/branch-month). Scale: 1,000 projects (raisable on request), 25 branches included.

How other Postgres providers compare on REST APIs

Most managed Postgres providers expose a REST API for cluster lifecycle, but they differ in granularity and provisioning latency.

AWS (RDS / Aurora) uses the AWS API (and SDKs / CloudFormation / Terraform) for CreateDBInstance and CreateDBCluster. The API call returns quickly but the database isn't usable until the instance reaches available state, which takes minutes. Per-PR or per-CI workflows aren't a great fit because billing is per instance-hour.
Supabase Management API exposes POST /v1/projects to create a new project programmatically, and POST /v1/projects/{ref}/branches to create preview branches. Project creation includes the full Supabase stack (database, auth, storage, edge functions) and the deploy workflow waits for health checks. See Supabase Management API.
DigitalOcean Managed Databases has a REST API for creating clusters, but provisioning is also per-cluster (minutes) and there's no copy-on-write branching primitive.

Neon's distinction is that the POST /branches endpoint returns a working connection string in seconds because branches are a metadata pointer to existing storage, not a physical clone. That's what makes per-PR, per-CI, and per-tenant flows practical.