← All error guides

Claude “Overloaded” error (529): what it means and how to fix it

The message you're seeing

• HTTP 529
• "type": "overloaded_error"

Is it you, or is Claude having problems?

A 529 always comes from Anthropic's side. Short bursts of them at busy times are normal and usually clear within minutes, even when there's no official incident. If they last a long time, line up with a problem on Anthropic's status page, or our verdict above shows problems, it's a wider outage and the only real fix is to wait.

Don't confuse it with similar-looking errors, which mean different things and need different fixes:

• 429 rate_limit_error: your organisation has hit a rate limit or spend cap. Anthropic notes that a sharp jump in your own usage can also cause 429s, so ramp traffic up gradually.
• 500 api_error: an unexpected internal error at Anthropic. Retry with backoff; if it keeps happening, contact support with the request ID.
• 504 timeout_error: the request took too long. Use streaming for long responses.
• 400 invalid_request_error: something in your request is wrong, or a spend limit you set has been reached. Retrying won't help.

How to fix it in your app

1. Retry with exponential backoffWait a moment, then retry, doubling the wait each time (for example 1, 2, 4, 8 seconds) and adding a little randomness so many clients don't retry at the same instant. Stop after a sensible number of attempts.
2. Let the official SDK helpAnthropic's SDKs already retry transient failures, including 5xx errors, twice by default with exponential backoff, and they respect the retry-after header when it's present. You can raise the max_retries setting if your app can afford to wait longer.
3. Don't make it worseRetrying instantly in a tight loop adds load at exactly the wrong moment. Queue work and spread retries out instead.
4. Move non-urgent work to batchesFor jobs that don't need an instant answer, the Message Batches API lets you submit requests and collect results later, instead of holding a connection open while the API is busy.
5. Plan a fallbackIf your product can't wait, decide in advance what happens during a long overload: a different Claude model, another provider, or a friendly “try again shortly” message.
6. Keep the request IDEvery Claude API response includes a request-id header. Log it, so you can give it to Anthropic support if errors continue.

Seeing “overloaded” in claude.ai or Claude Code?

The Claude apps and Claude Code run on the same infrastructure, so at busy times they can show a message that Claude is overloaded or couldn't finish a reply, and Claude Code may print the same 529 overloaded_error text. Wait a minute and try again.

This is different from hitting your plan's usage limit, which tells you when your limit resets. See our guide to Claude's usage limit if that's the message you're getting.

Questions

How long do Claude overloaded errors last?

Usually minutes. If 529s continue for a long time, check the live status above and Anthropic's status page: a longer run normally means a wider incident.

Is a 529 the same as a 429?

No. A 529 means the API is busy for everyone. A 429 means your own organisation hit a limit, so the fix is in your usage or account settings, not in waiting for Anthropic.

Should I switch to another model when Claude is overloaded?

If your app allows it, a fallback keeps things working during long overload periods. For short bursts, retrying with backoff is usually enough.

Related error guides

• Claude “usage limit reached”

Based on official documentation, checked 18 October 2026: Anthropic: Claude API errors. Error messages and limits change over time, so the provider's own docs are the final word.