OpenAI API error 429 rate limit exceeded

You're hitting OpenAI's API and getting slammed with error 429 — "Rate limit reached for requests." Your integration stops working, your users see failures, and you're stuck wondering if you need to upgrade or if somethi

OpenAI API error 429 rate limit exceeded

You're hitting OpenAI's API and getting slammed with error 429 — "Rate limit reached for requests." Your integration stops working, your users see failures, and you're stuck wondering if you need to upgrade or if something's broken.

What's actually happening

Error 429 means you've exceeded OpenAI's rate limits for your API tier. OpenAI restricts how many requests per minute (RPM), tokens per minute (TPM), and requests per day (RPD) you can send based on your usage tier and payment history.

Here's the exact error you'll see:

```

{

"error": {

"message": "Rate limit reached for requests",

"type": "requests",

"param": null,

"code": "rate_limit_exceeded"

}

}

```

The `type` field tells you what limit you hit — usually `requests` (too many API calls per minute) or `tokens` (too many tokens processed per minute). New accounts start on Tier 1 with strict limits: 500 RPM and 200,000 TPM for GPT-4o, for example. If you're processing large documents or running chatbots, you'll hit these fast.

Your tier automatically increases as you spend more and maintain your account in good standing. Check your current limits at platform.openai.com/settings/organization/limits — you'll see your exact RPM, TPM and RPD allowances for each model.

How to fix it

1. Check which limit you actually hit

Log into platform.openai.com/settings/organization/limits and look at your current usage versus your limits. If you're at 498/500 requests, you know immediately it's the request count. If tokens are maxed out, you're sending too much text per minute.

2. Add exponential backoff to your code

When you get a 429, wait before retrying. Start with a 1-second delay, then double it each time: 1s, 2s, 4s, 8s. Most official SDKs handle this automatically, but if you're using raw HTTP requests, you need to implement it yourself:

```python

import time

import openai

def call_with_backoff():

max_retries = 5

for i in range(max_retries):

try:

response = openai.ChatCompletion.create(...)

return response

except openai.error.RateLimitError:

wait_time = 2 ** i

time.sleep(wait_time)

raise Exception("Max retries exceeded")

```

3. Batch your requests

If you're making 50 separate API calls in quick succession, you'll hit request limits fast. Instead, combine multiple questions into a single API call or use the Batch API at platform.openai.com/batches. Batching costs 50% less and has no rate limits, though results come back asynchronously within 24 hours.

4. Request a tier increase

If you legitimately need higher limits, OpenAI automatically promotes you as you spend. Tier 2 requires $50 in successful payments and 7 days of account age. Tier 3 needs $100 and 7 days. Tier 4 requires $250 and 14 days. There's no manual application — just use the API and pay invoices on time.

Check how to contact OpenAI support if you believe your tier should be higher based on your spending but hasn't updated after 24 hours.

If that doesn't work

Error 429 sometimes appears alongside other OpenAI API errors like `insufficient_quota` or authentication failures. If you're seeing 429 but your usage dashboard shows you're nowhere near your limits, check:

  • Billing issues: Go to platform.openai.com/settings/organization/billing and verify your payment method isn't declined. A failed payment can trigger rate limits even if you're under quota.
  • Organization vs project limits: OpenAI now uses project-based rate limits. Check platform.openai.com/settings/organization/projects to see if your specific project has separate, lower limits.
  • Model-specific limits: GPT-4 has different limits than GPT-3.5. If you switched models recently, you might be hitting the new model's lower allowance.

If you're still stuck after checking all of this, contact OpenAI through platform.openai.com/account/support. Include your organization ID, the exact timestamp of the 429 error, and your current tier. Response time typically runs 2-3 business days.

Questions people actually ask

Q: How long does a rate limit last?

A: Rate limits reset every minute for RPM and TPM. If you hit 500 requests at 2:00 PM, you'll have a fresh 500 requests at 2:01 PM. Daily limits reset at midnight UTC.

Q: Will upgrading to a paid plan fix this?

A: There's no separate "paid plan" for higher limits. Your tier increases automatically based on total spending. Paying $50 moves you from Tier 1 to Tier 2, which roughly doubles your limits.

Q: Can I pay extra for custom rate limits?

A: Not through the standard API. Enterprise customers can negotiate custom limits, but that requires contacting OpenAI's sales team and typically involves contracts starting at $100k+ annually.

What to remember

  • Error 429 means you've exceeded requests per minute, tokens per minute, or requests per day
  • Check platform.openai.com/settings/organization/limits to see your exact allowances
  • Implement exponential backoff in your code — wait 1s, 2s, 4s, 8s before retrying
  • Your tier automatically increases as you spend: $50 for Tier 2, $100 for Tier 3, $250 for Tier 4
  • Use the Batch API for bulk processing to bypass rate limits entirely

---

*openai-support.com is an independent resource, not affiliated with OpenAI Inc.*

Related help