How much RAM does n8n need?

Short answer: 2 GB to start, 4 GB to be comfortable. n8n's own docs give 320 MB to 2 GB as an example range, and an idle n8n Cloud instance uses about 100 MB. On a VPS that also runs the OS and Docker, plan on 2 GB minimum for light personal use and 4 GB if you add Postgres, AI agent workflows or the Code node on large data.

What actually eats memory

• Data size per execution — big JSON responses and binary files (PDFs, images) sit in memory while a workflow runs.
• The Code node makes a copy of the data before and after it runs.
• Manual test runs — n8n keeps an extra copy for the editor.
• Concurrency — several workflows at once add up.
• Extras on the same box — Postgres, Redis (queue mode) and a reverse proxy each need their own RAM.

Signs you’ve run out

“Execution stopped at this node (n8n may have run out of memory)”, “Connection Lost”, 503 errors, or JavaScript heap out of memory in the logs. The Docker image restarts n8n on its own; an npm install may need a manual restart.

Try these before paying for more RAM

• Process data in batches (Loop Over Items + a sub-workflow that returns only a small result).
• Avoid the Code node for heavy lifting where a built-in node works.
• Don’t test huge datasets with manual runs.
• Raise Node’s heap with NODE_OPTIONS=--max-old-space-size=… and add a swap file as a buffer.

Next: Can n8n run on 2 GB? · Cheapest VPS for n8n · Calculator (it picks a RAM size from your execution count).

Sources: n8n docs — Fix memory issues, sizing prerequisites.