Traces#
Open Analytics → Traces to inspect individual AI requests end to end. Use it to debug latency, failures, routing, and payloads when Dashboards or Usage and Cost only show the aggregate picture.
Use this page when you need to:
- Investigate latency spikes, failures, or unexpected model behavior.
- Confirm which provider, model, routing rule, or virtual key handled a request.
- Audit who used which models and reconcile tokens or cost to a single call.
- Inspect prompt and completion payloads after routing or policy changes.
From this page you can learn success rate and volume for the selected filters, then drill into latency, tokens, cost, routing, and payloads for an individual request.
Next: filter by provider, model, or employee and open a row for trace details. Compare totals with Usage and Cost, or verify routing changes from Providers. For training-run telemetry, see Experiment Tracking — Tasks.
Trace details#
The detail panel shows the trace UUID, Provider, Model, Status, and Time, plus request metadata, metrics, captured payloads, and request parameters.
The Request details section displays request metadata, including the provider that handled the request, the full model identifier, the virtual key used for the request, the applied routing rule, the initiating employee or principal, and whether streaming was enabled.
The Metrics section displays request statistics, including total token usage with prompt and completion breakdowns, end-to-end latency, request cost, and the request timestamp.
The Request & Response section displays captured prompt and completion payloads. When request bodies are unavailable or not stored for the deployment, the panel displays No data.
The Params section contains the JSON request payload sent to the provider, including request parameters such as tools, function definitions, and parameter schemas. Use the copy control in the panel corner to copy the full payload for debugging or support purposes.