Engineering Workspace

Usage & Billing

Every inference request is metered. Query aggregated summaries, raw events, and a month-to-date billing snapshot.

Usage summary

GET/api/v1/usage/summary
ParameterTypeDescription
windowstringTime window: 7d, 30d, or month.
group_bystringAggregation key: provider, model, modality, or day.

Every usage event and trace carries a modality field from a closed set — text, vision, embedding, tts, stt, voice_realtime. Group with group_by=modality to split spend across text, image, embeddings, and voice surfaces.

CLI
stimulir usage --window 30d --group-by model
stimulir usage --window 30d --group-by model --json   # raw API response

Usage events

GET/api/v1/usage/events

Raw per-request usage events for the workspace — useful for reconciliation and debugging spend.

Billing snapshot

GET/api/v1/billing/hybrie/snapshot

A month-to-date rollup for the workspace, including:

  • MTD tokens and cost
  • By mode — mi (Managed Inference), byok, byoc
  • By provider
  • Daily series for charting
  • Adapter coverage — how much traffic is served with adapters

From the CLI

bash
stimulir usage --window month --group-by provider
stimulir billing snapshot