OpenAI API 429 rate limit error fix

You're making API calls and suddenly hit error 429 — "Rate limit reached for requests." Your requests stop working, your application stalls, and you're not sure why OpenAI is blocking you when you're paying for access.

OpenAI API 429 rate limit error fix

You're making API calls and suddenly hit error 429 — "Rate limit reached for requests." Your requests stop working, your application stalls, and you're not sure why OpenAI is blocking you when you're paying for access.

What's actually happening

Error 429 means you've exceeded OpenAI's rate limits — the maximum number of requests you can make in a given timeframe. OpenAI sets these limits based on your usage tier, which depends on how much you've spent historically on their API.

The actual error message looks like this:

```

{

"error": {

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

"type": "requests",

"param": null,

"code": "rate_limit_exceeded"

}

}

```

You'll see variations specifying whether you hit the requests-per-minute (RPM) limit, tokens-per-minute (TPM) limit, or requests-per-day (RPD) limit. New accounts start at the lowest tier with tight restrictions — often just 3 requests per minute for GPT-4 models. If you're building anything that makes multiple calls in quick succession, you'll hit this wall immediately.

This isn't about your total account balance. You can have $100 in credits and still get rate limited because OpenAI calculates limits based on cumulative spend over time, not current balance.

How to fix it

1. Check your current rate limits

Go to https://platform.openai.com/settings/organization/limits. You'll see your exact limits for each model — requests per minute, tokens per minute, and requests per day. If you're on the free tier or have spent less than $5 total, your GPT-4 limit is probably 3 RPM and 40,000 TPM.

2. Implement exponential backoff in your code

When you get a 429 error, wait before retrying. The response includes a `Retry-After` header telling you how many seconds to wait. Here's the pattern:

  • First retry: wait 1 second
  • Second retry: wait 2 seconds
  • Third retry: wait 4 seconds
  • Continue doubling up to a maximum wait time

Most API libraries support this automatically. In Python with the `openai` package, version 1.0+ retries 429 errors by default with exponential backoff.

3. Add at least $5 to your account and make successful API calls

Your usage tier increases as you spend more. The tiers are:

  • Tier 1: $5+ spent → 500 RPM for GPT-4
  • Tier 2: $50+ spent → 5,000 RPM
  • Tier 3: $100+ spent → 10,000 RPM
  • Tier 4: $250+ spent → higher limits
  • Tier 5: $1,000+ spent → highest standard limits

Add credits at https://platform.openai.com/settings/organization/billing/overview. Your tier won't upgrade immediately — you need to actually spend the threshold amount through successful API calls over time, and upgrades can take 24-48 hours after hitting the spend threshold.

4. Batch your requests if possible

Instead of making 20 separate API calls in one second, implement a queue system that spaces requests out. Use the Batch API (https://platform.openai.com/docs/api-reference/batch) for non-urgent processing — it's 50% cheaper and exempt from rate limits.

5. Request a rate limit increase

If your usage is legitimate and you're already at a higher tier, go to https://platform.openai.com/settings/organization/limits and click "Request increase" next to the limit you're hitting. Explain your use case specifically — "building customer service chatbot expecting 1000 daily users" works better than "need more requests." Approval takes 3-7 days and isn't guaranteed.

If that doesn't work

Check if you're hitting other API errors beyond 429 — 401 authentication failures and 500 server errors need different fixes. If you've added funds and waited 48 hours but your tier hasn't increased, contact OpenAI support with your organization ID and transaction receipts showing the spend. Include specific dates and amounts.

If you're getting charged more than expected while dealing with rate limits, see our guide on unexpected OpenAI charges.

Questions people actually ask

Q: I added $100 but still get 429 errors — why?

A: Adding credits doesn't immediately raise limits. You need to actually spend that amount through API usage over time, then wait up to 48 hours for the tier upgrade to process.

Q: Can I pay extra to skip rate limits completely?

A: No. Even enterprise accounts have limits. You can request increases through the limits page, but OpenAI reviews each request individually.

Q: Does error 429 consume my credits?

A: No. Failed requests don't count toward your bill. Only successful API calls that return completions are charged.

Q: My tier shows 500 RPM but I still hit limits at 200 requests — what gives?

A: You might be hitting TPM (tokens per minute) instead of RPM. A single request using 100,000 tokens counts more toward your TPM limit than 10 short requests of 200 tokens each.

What to remember

  • Rate limits are based on cumulative spending, not current account balance
  • New accounts start with severe restrictions — 3 RPM for GPT-4
  • Always implement exponential backoff with retry logic in your code
  • Tier upgrades take 24-48 hours after reaching the spend threshold
  • The Batch API bypasses rate limits entirely for non-urgent work

---

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

Related help