// WHAT IT MEANS
The upstream model provider — Anthropic, OpenAI or Google — is rate limiting the requests Cursor sends on your behalf. Your account is fine and your monthly pool is not involved; the third-party pool that the request would have drawn from is not charged for a rejected call.
It clusters around launches and outages at the provider, and around peak US hours.
// WHEN IT RESETS
Typically minutes, occasionally longer during a provider incident. There is no reset time on your side because nothing of yours was consumed.
If the provider is having a real outage, its status page (status.anthropic.com, status.openai.com) will show it; Anthropic and OpenAI have both reset their own subscribers' limits after large incidents, though that does not apply to API traffic through Cursor.
// WHAT TO DO NOW
- 01 Switch to a different model — another provider's frontier model, or a Cursor first-party model — and continue; the throttle is per provider.
- 02 Retry the same model after a few minutes if you need it specifically.
- 03 For long agent tasks, enable a fallback model in Cursor settings so the run does not stall when one provider throttles.
- 04 Check the provider's status page before assuming it is a Cursor problem.
// DOES IT AFFECT MY WEEKLY ALLOWANCE?
No allowance is spent: rejected calls are not billed against the third-party pool. Cursor has no weekly limit at all — the monthly pool resets on your billing date.
// ALSO APPEARS AS
- “Rate limit with provider”
- “Provider rate limited”
- “This model provider is currently rate limited”