This onboarding flow configures telemetry activation so the platform can compute unit economics across your models, tenants, and workload classes.
Confirm deployment surface and fleet scale. We configure telemetry activation and baseline computation.
Start ActivationFleet-wide baseline for tokens/sec/$, fairness distribution, and leakage signals — ready for policy enablement.
Common signals: KV-cache churn, decode stalls, fragmentation, CPU bottlenecks, batching inefficiencies, and cross-tenant interference.