Skip to main content

​
Observability

LLMGrid provides full observability for all requests across tenants and projects. This includes real-time dashboards, structured logs, metrics, and audit events.

​
Key Features

  • Traces
    Every request gets a unique x_llmgrid_trace_id for correlation across logs and dashboards.
  • Metrics
    • Latency per provider/model
    • Token usage per user/team/project
    • Error rates and fallback activations
    • Cost projections based on configured pricing
  • Dashboards
    View usage and performance in the LLMGrid Console:
    https://app.llmgrid.ai/observability
  • Audit Logs
    Track admin actions, route changes, and policy updates with actor identity and timestamp.

​
Example Trace Object

{
  "trace_id": "trace_abc123",
  "tenant_id": "TENANT_123",
  "project_id": "PROJ_abc",
  "route": "chat_default",
  "provider": "openai",
  "model": "gpt-4o-mini",
  "latency_ms": 850,
  "retry_count": 1,
  "tokens": { "prompt": 120, "completion": 210 }
}

​
Webhooks & Export

Send audit and usage events to external systems: 
POST /v1/admin/webhooks
{
  "url": "https://siem.yourdomain.com/llmgrid",
  "events": ["audit", "metrics", "route_change"]
}

​
BI & SIEM Integration

Export usage data to Power BI, Looker, or Tableau. Push logs to Splunk, Elastic, or Azure Monitor for compliance.

​
Redaction & Sampling

Enable prompt/response redaction for sensitive data. Configure sampling to control log volume in high-throughput environments.