API rate limits

Most APIs have a limit of 5000 per minute.

The following rate limits apply to specific API endpoints.

Note: These limits are enforced per IP address and tenant. The admin/users limit is also counted per API token.

Duration-based limits

Separately from the request counts above, Tines limits how much server time slow API requests can use. This protects every tenant on the platform when requests slow down, so you can get a 429 well below the requests-per-minute limits.

Requests that finish within 5 seconds don't use up any of the budget. Once a request runs longer than 5 seconds, its full duration counts against a budget of 180 seconds of request time per 30-second window.

Budgets are tracked per IP address and tenant, regardless of API token. Case requests that read data (GET), case requests that change data, and all other API requests each have their own budget, so slow requests to one endpoint can cause 429 responses on other endpoints that share its budget.

Record list requests (GET /api/v1/records and POST /api/v2/records/search) also share one budget across your whole tenant, regardless of IP address. Record query requests (POST /api/v1/records/query and POST /api/v2/records/aggregate) share another.

When a budget runs out, every new request it covers gets a 429 response that says "Slow requests detected", including requests that would have finished quickly, until the window resets.

Handling 429 responses

The response body tells you which kind of limit you hit. Request-count limits say "Too many requests are being made to this host" and include a Retry-After header with the number of seconds to wait. Duration-based limits say "Slow requests detected" and don't include Retry-After.

In both cases, retry the request after a short delay. In an HTTP Request action, add 429 to Retry on status. Under normal conditions, API requests finish well within the duration-based limits, so if you see "Slow requests detected" often, contact support.