How to Fix ChatGPT Error 429: Too Many Requests (2026)

How to fix ChatGPT error 429 starts with understanding one thing: 429 means “too many requests.” ChatGPT, or the OpenAI API behind it, is telling you that you have hit a limit on how many requests you can send in a short time. It is a temporary block, not a broken account, and in most cases it clears on its own if you stop hammering the send button.

If ChatGPT has stopped working in other ways too, the general fixes in our guide to how to fix ChatGPT not working are a good first stop. Error 429 is a specific case, though, so it has its own fixes.

What error 429 actually means

HTTP 429 is a standard web status code. A server sends it when a client sends more requests than the server will accept in a given window. With ChatGPT, that window can be a few minutes or an hour, depending on your plan and which model or tool you are using.

There are three common versions of the problem, and the fix depends on which one you have:

• ChatGPT app or website limit. You see “you’ve reached your limit” or a reset time. Your plan allows a set number of messages with a model, and you have used them up for now.
• Tool limit. Image generation, file uploads, or data analysis can have separate limits from plain chat. One tool can be blocked while normal chat still works.
• API rate limit. If you build with the OpenAI API, your project has requests-per-minute and tokens-per-minute limits. Hitting either one returns a 429 with details in the response.

Fix 1: Stop, wait, and read the message

The most effective fix is also the least satisfying: stop sending messages. Every extra send or regenerate click is another request, and it can extend the wait. Read the exact message first. If it gives a reset time, wait until then. If it names a model or a tool, that tells you what is limited.

As a rough guide, wait at least a few minutes for a short throttle, and until the stated reset time for a plan limit. Refreshing the page every 30 seconds does not speed this up.

Fix 2: Check OpenAI’s status page

Before you change any settings, check OpenAI’s status page. If there is an incident affecting ChatGPT, waiting is the only fix, and no amount of clearing cache will help. This step takes 30 seconds and saves a lot of wasted troubleshooting.

Fix 3: Close extra sessions and tools

Open ChatGPT in one place at a time while you are limited. Close extra browser tabs, the phone app, and the desktop app if they are all open, and stop any long-running task or upload. Each open session can add requests, and a stuck reply still counts as work in progress. If a reply looks stuck, wait 30 to 60 seconds, stop it once, and then try again once, not repeatedly.

This is also the fix that helps when the error looks like a ChatGPT not loading problem: too many overlapping sessions can make the app feel frozen when the real cause is the limit.

Fix 4: Try a different model or tool

Limits are often per model. If GPT-5 is limited, a lighter model may still be available on your plan. The same goes for tools: if image generation is blocked, plain text chat may still work. Switching models is not a workaround to abuse; it is simply using the allowance you still have.

If you are on the free plan and hit limits often, that is normal. Free plans have the tightest limits by design, and heavy use will hit them every day.

Fix 5: Clear the browser basics, once

If the error follows you after waiting, try the standard browser cleanup once: sign out, clear ChatGPT’s site data or cookies, close the browser, and sign back in. Also try turning off a VPN or a privacy extension for a test, since shared VPN addresses can hit limits faster because many people share one address. If none of this changes anything, the limit is on your account or plan, not your browser, so stop repeating these steps.

Fix 6: For API users, slow down properly

If you get 429 from the API, the fix is in your code, not your browser:

• Honor the Retry-After value when the response includes one. That is the server telling you exactly how long to wait.
• Use exponential backoff. Wait a short time, then double the wait after each failure, with a little random jitter so many clients do not retry at the same second.
• Reduce concurrency. Send fewer requests at the same time, and space out bursts.
• Check your usage and budget limits. A 429 can also mean a spending cap or an exhausted balance, and no retry fixes a billing limit. Check your project’s usage page before adding more retries.

Remember that a ChatGPT subscription does not raise API limits. The two products have separate limits and separate billing.

What does not fix error 429

• Repeatedly refreshing the page or resending the same prompt.
• Creating extra accounts to dodge a limit, which can get accounts flagged.
• Clearing cache ten times. If it did not work the first time, the limit is elsewhere.
• Paying for an upgrade mid-throttle and expecting instant relief. Plan changes can take time to apply, and existing limits may still need to reset.

How to avoid error 429 next time

Space out long sessions instead of sending many prompts in a burst. Break big jobs into smaller chunks, and let each reply finish before sending the next prompt. If you use the API, build backoff and request queuing in from the start rather than adding them after the first 429. And if you hit the same limit every day at the same time, that is a sign your plan no longer matches your usage, not a bug to work around.

Last updated: October 2026

Frequently asked questions

How long does ChatGPT error 429 last?

It depends on the limit. A short throttle can clear in minutes. A plan message limit usually clears at the reset time shown in the message, which can be hours away.

Is error 429 the same as “ChatGPT is at capacity”?

Not exactly. Capacity messages mean the service is busy for everyone. Error 429 means your requests specifically have crossed a limit, even when the service is fine for others.

Will upgrading to Plus stop error 429?

It raises your limits, so you will hit them less often, but Plus still has limits. Heavy use can still trigger a 429 on a paid plan.

Should I contact OpenAI support about error 429?

Only if the error follows you across browsers, devices, and networks after the reset time has passed, or if you suspect a billing limit is wrong. For a normal throttle, support cannot lift the limit early.