All Collections
Claude API
Our approach to API rate limits
Our approach to API rate limits
Updated over a week ago

We currently measure API rate limits in terms of concurrent requests. Depending on your requests per minute, tokens per request, and model selection, one request can handle a range of traffic. For this reason, we try to be flexible with users to adjust their limit to accommodate their growth.

For API users in the non-commercial evaluation phase, we limit users to 1 concurrent request.

Once you sign our standard commercial contract, we automatically increase your API keys to serve up to 5 concurrent requests.

To increase your rate limit further, we require that you pass our Trust and Safety Review. You can request an increase by emailing with your estimated requests per minute and your estimated tokens.

Did this answer your question?