Réponse directe
Cette page explique comment les équipes utilisent la passerelle compatible OpenAI de NextModel. Audio SKUs today, and why this page ships no audio snippet yet. Elle ajoute les étapes pratiques, les points de configuration et les questions fréquentes.
Audio models today
Text-to-speech is live: POST /v1/audio/speech with a speech model from GET /v1/models (currently the openai/tts-1 family), billed per 10k input characters. Audio transcription routes exist but are not open yet on this deployment — a call returns 404 transcription_ingress_disabled until the flag is switched on. Audio SKUs are billed per 10k input characters; the dashboard playground covers audio trials. This page will carry tested TTS snippets the day the audio contract is verified against the live API — we do not publish untested code.