Skip to main content

Metrics & Usage Dashboard

The Usage dashboard provides real-time visibility into your API consumption, performance metrics, and spending.

Overview​

To access the dashboard, navigate to the Usage tab in your main dashboard.

Time Range Selector​

At the top right, you can filter all charts by a specific time period:

  • 24H: Last 24 hours (Day)
  • 7D: Last 7 days (Week)
  • 30D: Last 30 days (Month)

Available Charts​

1. Total Spend​

This chart displays your cumulative cost over the selected period.

  • Purpose: Track budget burn rate and identify cost spikes.
  • Data: Aggregated cost in USD across all models.

2. Requests & Tokens​

You can toggle this chart between two views to analyze volume:

  • Requests: The total number of API calls made.
  • Tokens: The total number of input (prompt) and output (completion) tokens processed.
  • Model Filtering: Use the dropdown to filter these metrics by a specific model (e.g., "Only show Mistral Large usage").

3. Latency Metrics​

Monitor the speed and responsiveness of your requests using two key metrics:

  • TTFT (Time to First Token): Measures how quickly the model starts generating a response. This is critical for perceived latency in chat applications.
  • Total Time: Measures the total duration of the request from start to finish.

Filtering by Model​

The Model Dropdown allows you to isolate metrics for individual models.

  • By default, charts show aggregated data for All Models.
  • Select a specific model to see its individual performance.

Best Practices​

  • Monitor TTFT: If you are building a real-time chat app, keep an eye on TTFT spikes.
  • Analyze Token Ratios: Check the Requests vs. Tokens view to understand if your average request size is increasing (which affects cost).
  • Cost Management: Use the Total Spend chart in conjunction with Billing Settings to manage your credits effectively.