My openAi node keeps returning 429 Too many requests

Questions
1

My Ai workflow works perfectly for a while then suddenly starts returning

Describe the problem/error/question

What is the error message (if any)?

Please share your

)

Chat trigger

Ai Agent

Opening chat model

Share the output returned by the last node

Information on your n8n setup

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
2

Hi @Oluwabunmi_Miracle_A

You have most likely run into rate-limit or run out of credits.

3

Hi mate,

This message : too many requests appears when you are out of plan, your API key has limitations and when you use more than expected pops up . check your API key validation.

4

“Works for a while, then suddenly 429” usually comes down to one of two different things, and the fix is different for each — worth 30 seconds to check which one you have:

1. Open the failed execution and read the actual error body.

  • If it says something like “You exceeded your current quota” (insufficient_quota) → you’ve run out of credits / hit your monthly budget cap. No retry will help; add credits or raise the budget limit on platform.openai.com.
  • If it says “Rate limit reached for … requests per min” or “tokens per min” → you’re bursting past your tier’s RPM/TPM. This is the classic “fine for a while, then 429” pattern, because loops send many requests in a tight burst.

2. If it’s the rate-limit flavor, two n8n-side fixes usually clear it:

  • On the OpenAI node, open Settings → Retry On Fail, set Max Tries 3–5 and Wait Between Tries ≥ 5000 ms. 429s are transient, so a short backoff often rides out the per-minute window.
  • If you’re processing many items in one run, put a Loop Over Items (Split in Batches) node with a small batch size and a Wait node (10–20 s) inside the loop, so requests are spread instead of fired all at once. TPM counts tokens, not just calls — long prompts hit the ceiling faster, so trimming the prompt or max_tokens also buys headroom.

Your account’s actual RPM/TPM limits are on platform.openai.com → Settings → Limits (new accounts start on low tiers, and limits rise with usage history).

If you can share the exact error text from the execution log, it’ll be obvious which case you’re in.

5

Hi @Oluwabunmi_Miracle_A
Check which credential your OpenAI Chat Model is set to. If it’s the free “100 OpenAI credits” one that n8n offers (it appears in your credential list with no API key you can edit), that’s a small shared allowance: the workflow runs fine until it’s used up or throttled, then every call comes back 429. That credit sits outside your own OpenAI account, so it can’t be raised from your own OpenAI billing.
Create your own OpenAI API key with billing active and select that credential on the node instead: