Try n8n with a free web service and free Postgres. Move both to paid instances before relying on schedules, webhook response times, or long-term workflow storage.
Each instance has a fixed monthly price, prorated by the second.
Capability | Render | |||
|---|---|---|---|---|
How n8n runs | Image-backed web service with Postgres | Long-running containers | Long-running dynos | Docker image on Fly Machines |
Free tier | Free for testing. Schedules need paid compute | $5 trial for 30 days, then $1/month credit | None. Eco: $5/month for 1,000 hours | No free tier. Personal bills under $5 waived |
Always-on, 2 GB | $25/month (1 CPU) | Up to $40/month at full use (1 vCPU) | $250/month (Performance-M, 2.5 GB dedicated) | $13.39/month shared CPU, $33 dedicated |
Postgres for workflow data | Render Postgres from $6/month, with PITR | Maintained Postgres template | Heroku Postgres. No rollback on Essential | Managed Postgres from $38/month, HA included |
Queue-mode workers | Workers plus Key Value as the queue | Workers plus a Redis template | Worker dynos plus Key-Value Store | Process groups plus Upstash Redis |
Persistent disks | SSDs that grow with no downtime (single instance) | Volumes (no replicas) | None. Files reset when dynos restart | Volumes, one per Machine, not replicated |
Stack as code | One render.yaml, auto-synced on push | Project file, applied with CLI or CI | app.json per app, read only at creation | fly.toml per app. No database definitions |
Competitor details as of November 2026. Instance prices are compute only. Fly.io prices are for US East.
Keep workflow data in Postgres and configuration in a render.yaml Blueprint. You maintain n8n versions and credentials.
Store workflows and execution records outside the n8n container. Choose paid database compute for ongoing use.
Keep a stable N8N_ENCRYPTION_KEY in your service environment and back it up securely. The database alone does not preserve the key.
Connect the main service, workers, and datastores in the same workspace and region using internal connection details.
Add n8n queue mode when you need it, with Postgres, a Redis-compatible queue, and separately billed workers.
Choose the n8n image version and deploy after testing. Image-backed services need a deployment trigger to pull updates.
Increase memory as workflow size and concurrency grow. Test binary-data storage and execution retention for your workload.
Create your own repository from the official Render n8n example, which defines an image-backed web service and Postgres database.
Create a Blueprint from your repository. Keep the generated encryption key, then confirm the service and database plans before deploying.
Open the service URL, create your n8n account, and configure its public webhook URL. Test a workflow before connecting production systems.