OverviewLive FeedIncidentsSLOsTracesReplay StudioLineageTopologyAgentsFleets

Platform

Model & Cost Intelligence

Model and provider latency, failure, token, and cost analytics with drift and anomaly detection.

Designed · pending wiring

Planned capabilities

  • Latency, failure, token, and cost by model / provider
  • Prompt / template drift
  • Unsupported-claim flags when available
  • Judge / model comparison
  • Cache-hit and retry rates
  • Spend by agent / org / fleet / tool / environment
  • Anomaly detection for cost and latency

Engineering contract

Route
/models
Data model
ModelCall { id, provider, model, tokens, cost, latency, outcome }
Primary actions
Compare modelsSet cost alertInspect drift