Usage & billing

Two separate usage rails

Nairon meters two distinct products separately rather than trying to reconcile them into one number:

• Voice calls (phone/web agents built with flows) are metered by call duration, recorded directly on the call record at call time.
• API-product usage (platform-api’s chat/speech/transcription/realtime/batch/KB calls) is metered per invocation — seconds of audio, characters synthesized, tokens in/out — stamped with the organization and project the API key belongs to.

If you’re building hotline flows, see the Client guide for how calls are recorded and shown in analytics. This page covers API-product usage.

Reading usage

• In the dashboard: a platform-type project’s console on platform.v2.nairon.xyz shows usage broken down by service (chat/speech/transcription/realtime/batch/KB) over time, plus an invoice preview.
• Programmatically: GET /v1/projects/{projectId}/usage and GET /v1/orgs/{orgId}/usage on service-api, or the equivalent get_usage MCP tool.

How billing numbers are produced

Every platform-api invocation is recorded into a usage-event log and rolled up over time into per-organization, per-project, per-service usage totals for a given time window. Those rollups — not the raw event log — are what invoice previews and billing are computed from, since the raw per-invocation log is a best-effort, high-volume analytics stream rather than an exactly-once billing ledger. Rate cards (price per unit, per service/metric) determine how a usage quantity translates into an invoice-preview amount; organization-level overrides exist for headless integrators billing their own tenants differently from the standard rate card.