Response time
Average time to a successful answer, seconds
Data refreshes every minute
Model statistics
Average response time and the share of requests that failed on our side, for every API429 model over the last 6 hours. No request volumes or customer data.
Errors
Share of requests that failed on our side, %
Pick up to 6 models to compare them on the charts.
How we measure
Response time is the average time from receiving a request to the complete answer, over successful requests. For streaming it is the time to the end of the stream; for images, video and voice, to the finished result.
Errors are the share of requests that failed on our side or at the provider: overload, 5xx failures, an exhausted pool. Errors in the request itself (invalid parameters, no balance, your own key limits) are not counted.
When a model had no requests in an interval, the line joins the neighbouring points. Models without requests in the last 6 hours are not listed.