Platform

Models and pricing

Choose Standard or Fast with Free or Partner access.
View as Markdown

All prices are in USD. Models ending in :free are free to use within their endpoint-specific usage limits. Partner models require early access and available credits.

Choose a class

ClassBest for
FastApplications that need sub-50 ms server-side time-to-first-audio (TTFA) for TTS or time-to-final-segment (TTFS) for STT
StandardExtremely low prices when higher server-side latency is acceptable

Text to Speech

ClassAccessModel IDPrice per 1M input characters
StandardFreeqwen3-tts:freeFree
FastFreeqwen3-tts-fast:freeFree
StandardPartnerqwen3-tts$5
FastPartnerqwen3-tts-fast$10

Set model in the speech generation request body.

Speech to Text

ClassAccessModel IDPrice per input audio hour
StandardFreeqwen3-asr:freeFree
FastFreeqwen3-asr-fast:freeFree
StandardPartnerqwen3-asr$0.06
FastPartnerqwen3-asr-fast$0.12

Set session.model in the initial session.configure when configuring a session.

Partner access

Partner endpoints provide high concurrency through early access during public alpha. If you are interested in early access, contact founders@narilabs.com. We are working to make these endpoints generally available after public alpha.