Usage and statistics
The console shows what your account has used, per day and per model, over the last 30 days.
Overview
The overview shows your balance, request count, token count and error rate for the last 30 days, plus daily charts and a breakdown by model with cost and average latency.
Statistics
Statistics charts average latency and failed requests per day. Latency is measured from receipt of the request to the first token, so streamed responses show their time to first token.
Accuracy
- Usage appears a few seconds after a request completes.
- Totals are computed from the same records that are billed, so usage and charges always match.
- Days are in UTC.
A day with no requests shows zero. The charts never estimate or fill in missing values.