Troubleshooting · Cursor

Cursor Rate Limit / 429 With Your Own Anthropic Key

With a custom Anthropic key, a 429 in Cursor is the key's own per-minute ceiling, not Cursor's plan limits. Why long chats trigger it and how to stop the loop.

What you see

HTTP 429
{"type":"error","error":{"type":"rate_limit_error","message":"This request would exceed your organization's rate limit of 80,000 input tokens per minute. Please reduce the prompt length or the maximum tokens requested, or try again later."}}

Why it happens

  • The per-minute token ceiling of the organization behind your key was exceeded. With a custom key, Cursor's own plan limits are out of the picture — this is your key's budget.
  • Long chats resend the whole conversation plus attached context every message, so one busy tab can spend a minute's token budget alone.
  • The same key is shared with other tools or teammates, and their traffic counts against the same window.
  • Rapid-fire retries after the first 429 keep the window full.

How to fix it

  • Start a new chat for a new task instead of extending one giant conversation — context resent per message is what eats the budget.
  • Give Cursor its own key if the current one is shared, so one tool's burst cannot starve another.
  • Wait out the minute window before retrying; tight manual retries prolong the condition.

Related searches

  • cursor rate limit exceeded custom api key
  • cursor 429 anthropic key

FAQ

Is this Cursor's fast-requests limit?+
No. With your own key, requests bypass Cursor's plan metering entirely. The 429 comes from the API organization behind your key.
Why do long conversations hit 429 more often?+
Every message resends the whole history and attachments. The visible reply is small, but the counted input tokens grow with the conversation.
Does upgrading my Cursor plan help?+
Not for this error. The ceiling belongs to the API key. Reduce per-minute usage, stop sharing the key, or arrange higher throughput with its issuer.

Skip the broken setup

apiToken.sale serves the standard Anthropic API — same models, same SDKs, one prepaid balance. Point your tool's base URL at it, and the config on this page works as written.