Skip to content

Token Metering ​

Aggregate the tokens and requests handled by endpoints over a period. Break usage down by project, user, model, or endpoint and download it as CSV for usage reviews or internal chargeback.

Prerequisites ​

  • An administrator account with access to AI Inference Admin

Click Evaluation → Token metering in the sidebar.

Period · Export ​

Choose the period to aggregate and download the results as CSV.

Period · Export
No.FeatureItemDescription
1-1PeriodLast 7d, etc.· Choose a preset period such as the last 7 days, or a custom range.
1-2Export CSVExport· Downloads the aggregated results for the current period and grouping as a CSV file.

Summary metrics ​

Total tokens and requests handled by endpoints in the selected period.

Summary metrics
No.FeatureItemDescription
2-1MetricsTotal tokensAll tokens, input and output combined.
Prompt tokensInput tokens received in requests.
Completion tokensOutput tokens generated by models.
RequestsNumber of requests handled.
Streamed requestsRequests answered with streaming.

Tokens over time ​

Daily token usage as bars.

Tokens over time
No.FeatureItemDescription
3-1Group byBar grouping· Choose whether bars are split by project, user, model, or endpoint.

Usage breakdown ​

View token usage by user, project, model, or endpoint in a table.

Usage breakdown
No.FeatureItemDescription
4-1TabsBy user / By project / By model / By endpoint· Choose how to break down the table. Use the search box to find an item.
4-2ColumnsPrompt · Completion · Total · Requests · % of total· Input, output, and total tokens, request count, and share of the total for each item. Click a project name to open it.

The New Standard for AI Infrastructure