n8n-io/n8n · error · OperationalError

OpenAI: Rate limit reached

Error message

OpenAI: Rate limit reached

What it means

When the OpenAI SDK throws a RateLimitError, the openAiFailedAttemptHandler intercepts it and rethrows as an OperationalError with a mapped message. The error code from the OpenAI response is looked up in errorMap (insufficient_quota, rate_limit_exceeded). If no specific code matches, the generic 'OpenAI: Rate limit reached' message is used. This is an operational, transient error — retryable.

Source

Thrown at packages/@n8n/nodes-langchain/nodes/vendors/OpenAi/helpers/error-handling.ts:50

		error.message.includes('not a chat model')
	);
}

export const openAiFailedAttemptHandler = (error: unknown) => {
	if (isNonChatModelError(error)) {
		throw new OperationalError(
			'This model requires the Responses API. Enable "Use Responses API" in the OpenAI Chat Model node options to use this model.',
			{ cause: error },
		);
	}

	if (error instanceof RateLimitError) {
		// If the error is a rate limit error, we want to handle it differently
		// because OpenAI has multiple different rate limit errors
		const errorCode = error?.code;
		const errorMessage =
			getCustomErrorMessage(errorCode ?? 'rate_limit_exceeded') ?? errorMap.rate_limit_exceeded;
		throw new OperationalError(errorMessage, { cause: error });
	}
};

View on GitHub (pinned to 5ac6606e81)

Solutions

1. Reduce request concurrency and add delays between calls to stay under RPM/TPM limits
2. Upgrade your OpenAI API usage tier to increase rate limits
3. If insufficient_quota, add billing credits or upgrade the plan at platform.openai.com/settings/billing
4. Implement exponential backoff retry at the workflow level using a Loop node with a Wait

Example fix

// before — firing 50 concurrent OpenAI requests
// after  — batch to 5 concurrent with 1s delay between batches

// Estimate token/request budget before calling OpenAI
const estimatedRPM = items.length; // requests per minute
const rpmLimit = 500; // your tier's RPM limit
if (estimatedRPM > rpmLimit * 0.8) {
  // throttle or batch
  items = batchItems(items, Math.floor(rpmLimit * 0.8));
}

import { RateLimitError } from 'openai';

function isOpenAiRateLimitError(error: unknown): error is RateLimitError {
  return error instanceof RateLimitError;
}

// In a custom node or code node wrapping OpenAI calls
try {
  await openAiCall();
} catch (error) {
  if (error instanceof RateLimitError) {
    const retryAfter = error.headers?.['retry-after'];
    await sleep(parseInt(retryAfter ?? '60') * 1000);
    return openAiCall(); // retry once
  }
  throw error;
}

When it happens

Trigger: The OpenAI account has exceeded its rate limits (requests-per-minute, tokens-per-minute) or has run out of quota (insufficient_quota code). The error comes from the openai SDK's RateLimitError class during any API call (chat, embeddings, images, etc.).

Common situations: Running parallel/bulk workflows against OpenAI; hitting the TPM limit on high-token payloads; account is on a free/tier-1 plan with low RPM; monthly quota exhausted.

Related errors

• Rate limit exceeded. Please wait a moment and try again.
• admittance_rejected
• Failed to retrieve API Key
• Database connection timed out
• Timed out after ${timeoutMs}ms waiting for database connecti

AI-assisted analysis of n8n-io/n8n@5ac6606e81 (2026-08-12). Data as JSON: /api/errors/b32cd30ce1e3078c. Report an issue: GitHub.