Skip to main content

AI Usage Tracking

Monitor and optimize your AI infrastructure.

Overview​

Tracking AI usage is essential for managing costs, identifying performance issues, and understanding how users are interacting with your AI features. LibreApps Desktop provides built-in tools for logging and analyzing AI requests.

Key Metrics​

  • Token Usage: Track the number of input and output tokens for each request.
  • Latency: Measure the time it takes for the AI to respond.
  • Cost: Calculate the estimated cost of each request based on the provider's pricing.
  • Success Rate: Monitor the percentage of successful vs. failed AI requests.
  • User Activity: Identify which users are using AI features most frequently.

How it Works​

The LiteLLM proxy automatically logs every request and response to a database (e.g., PostgreSQL or Redis). This data can then be visualized using tools like Grafana or accessed via a custom management dashboard.

Configuration​

To enable usage tracking in LiteLLM, update your config.yaml:

litellm_settings:
  success_callback: ["langfuse", "prometheus"]
  failure_callback: ["langfuse"]

In this example, LiteLLM is configured to send usage data to Langfuse (for detailed tracing) and Prometheus (for real-time monitoring).

Best Practices​

  • ✅ Do this: Set up alerts for unusual spikes in AI usage or costs.
  • ✅ Do this: Use usage data to identify opportunities for model optimization (e.g., switching to a cheaper model for certain tasks).
  • ❌ Don't do this: Ignore usage tracking until you receive a large bill from your AI provider.