Unrefused
Sign inGet an API key
Menu

Rate limits

Two limits per key: requests per minute and in-flight requests.

Both limits are per API key and configurable when you create the key. Defaults are 600 requests per minute and 16 concurrent requests.

Headers

code
x-ratelimit-limit-requests:     600
x-ratelimit-remaining-requests: 598
retry-after: 1                     (only on 429)

Behaviour

  • Requests per minute is a token bucket that refills continuously, so short bursts above the average are fine.
  • Concurrency counts requests in flight. A streaming response holds its slot until the stream closes, so long generations consume concurrency, not RPM.
  • Limits are shared across all gateway instances.
  • On 429, wait retry-after seconds. The OpenAI SDKs do this automatically.

Need more? Raise the per-key limits on the API keys page when you create a key, or split traffic across keys per workload.

Spot an error or a gap? Tell us — and quote x-request-id if you have one.