Observability and Usage Monitoring

Use Logs, Usage, and Audit to manage reliability and spend

Monitoring model operations end-to-end

Customer Dashboard combines request telemetry and billing visibility so teams can detect failures quickly and keep spend predictable.

Logs: how to use each tab

Metrics tab

Use for trend detection:

• Request volume changes
• Latency shifts
• Token growth patterns
• Geographic traffic movement

Requests tab

Use for deep investigation:

• Filter by model, endpoint type, token usage, and latency
• Open specific inference rows
• Validate payload and response details

Usage and Billing

• Track total requests, tokens, and cost over time.
• Compare by project and time range.
• Export for reporting and finance reviews.
• Configure billing alerts for proactive guardrails.

Audit for traceability

When diagnosing incidents, correlate request outcomes with audit events (key changes, project updates, or policy edits).

Incident response playbook

1. Detect anomaly in Metrics.
2. Locate affected requests in Requests tab.
3. Confirm impact in Usage.
4. Review related changes in Audit.
5. Apply fix and monitor recovery trend.