We’re currently running a live medical congress using LiveKit Cloud (Ship plan). We built a real-time transcription + translation tool: speaker audio → Deepgram STT → GPT-4o-mini translation → Cartesia TTS.
Question : Billing page refresh
Is the usage page on cloud.livekit.io updated in real-time or with a delay? We need to monitor our TTS consumption during the event and can’t tell if what we see is current.
Any insight appreciated, especially from people who’ve run similar broadcast/translation setups.
My recommendation is to capture the usage in your code and transmit that to your server for near real time monitoring.
You can get TTS usage from the session_usage_updated event or from session.usage. Each entry in session.usage.model_usage includes a TTSModelUsage object with fields like characters_count, audio_duration, and token usage.
Subscribe during the session to track usage live, or read session.usage in a shutdown callback for cumulative totals. See the Session usage and TTSModelUsage sections in Data hooks.