Skip to main content
The AI Gateway supports exporting data to OpenTelemetry (OTel) compatible collectors, enabling integration with your existing observability infrastructure. Choose the export type based on your needs:

Analytics

Export aggregated analytics data including request counts, latency metrics, token usage, and cost tracking to OTel-compatible collectors.

Complete Logs

Export complete LLM request/response logs with full prompt and completion content following GenAI semantic conventions.

​
Comparison

​
Use Cases

​
Analytics

Best for:
  • Building dashboards in Datadog, Grafana, or similar tools
  • Setting up alerts on error rates or latency
  • Long-term storage of usage metrics
  • Cost tracking and attribution

​
Complete Logs

Best for:
  • Debugging specific LLM interactions
  • Quality analysis of prompt/completion pairs
  • Integration with LangSmith or similar GenAI observability tools
  • Compliance requirements needing full audit trails

​
Getting Started

1

Choose your export type

Determine whether you need aggregated metrics or complete logs based on your use case
2

Configure environment variables

Set the appropriate environment variables for your chosen export type
3

Deploy and verify

Redeploy your AI Gateway and verify data appears in your OTel collector
Both export types can be enabled simultaneously if you need both aggregated metrics and complete logs.
Last modified on September 15, 2026