Metrics

The Metrics page provides detailed analytics about your prompt executions. You can visualize data for token usage, latency, cost, and duration over time.

Key features of the Metrics page include:

• Metric Tabs: Switch between different metrics:

  • Tokens: The number of tokens processed.
  • Latency: The time it took for the LLM to respond in milliseconds.
  • Cost: The cost of the LLM execution in USD.
  • Duration: The total time for the prompt execution to complete in milliseconds.

• Date Range Picker: Select a custom date range to view metrics for a specific period.

• Grouping and Filtering:

  • Group by: Group the data by prompt, version, or model to see how each contributes to the overall metrics.
  • Filters: Filter the data by specific prompts, prompt versions, or models to narrow down your analysis.

• Interactive Chart: The main chart visualizes the selected metric over the chosen time period. If you group the data, the chart will display a stacked area chart showing the breakdown by group.

• Data Export: Download the raw chart data as a CSV file for further analysis.

• Total Metrics: A card displays the total (or average, for duration) for the selected metric over the specified period.