Skip to content

Metering

Voice AI runs on a credit-based prepaid model. Every account is provisioned a Lago wallet on signup, seeded with free credits, and replenished by top-ups or plan renewals.

Authorisation against the wallet is enforced in real time at the gateway, backed by Redis with sub-millisecond latency. Calls without sufficient credit fail fast with 402 Payment Required rather than running and being billed retroactively.

Per-API measurement units

Each API has its own natural unit. Credits are debited at the published per-unit rate when the call completes successfully. Failed calls are not billed.

APIUnitNotes
Speech-to-TextSeconds of audio submittedRounded up to the nearest second per request.
Text-to-SpeechCharacters of input textWhitespace and punctuation included.
TranscriptionsSeconds of audio submittedSame rate as Speech-to-Text; diarization included.
Voice CloneRequestsOne credit-priced request per clone, irrespective of sample length.
Voice DesignRequestsOne credit-priced request per generated voice.
Speech TranslationSeconds of audio submittedIncludes translation; no separate text fee.
Text TranslationCharacters of source textOutput length not billed separately.
ReasoningTokens (input + output)Standard tokenizer; reported in the response body.
Speech-to-SpeechSeconds of audio submittedPatent-pending pricing tier.
Streaming TextTokens (input + output)Streaming has no surcharge over Reasoning.
Language IntelligenceRequestsSingle credit-priced request per call.
Knowledge & MemoryRequestsIndexing and retrieval each priced separately.

The current per-unit rate for your plan is shown on the Console pricing page.

Free credits

Every new account receives a free credit grant on signup. The grant is scoped to the Voice AI wallet and does not transfer to Universal API.