Deepgram triples concurrency limits for AI voice agents
Deepgram said it has raised default concurrency limits for Voice Agent API, Streaming STT and TTS across plans.
Voice AI specialist Deepgram has raised default concurrency limits for its voice products, tripling baseline capacity for many customers and offering higher ceilings on paid plans.
The change is intended to support systems - such as conversational AI platforms, meeting intelligence products and contact center applications - moving from pilot to production.
The company, in an announcement, said the new defaults rise to 45 concurrent connections for Pay as You Go voice agent APIs, up from 15, whilst customers on its Growth Plan will be able to 60, also up from 15.
Streaming STT connections rise to 150 on Pay as You Go 150, from 50, whilst the Growth Plan cap rises to 225 (from 50).
For Deepgram's WSS TTS streams it increases to 45 on Pay as You Go 45 (from 15) and 60 on Growth Plan 60 (from 15).
Related reading
- New guide ranks leading speech-to-text APIs for voice application developers
- Deepgram tool lets AI assistants answer phone calls
- Udemy to insert its courses directly into ChatGPT
Growth Plan customers can receive up to 4.5x concurrency compared with prior defaults, Deepgram highlighted.
Deepgram compared its updated offering to rival voice AI vendors, claiming other platforms use ramp-up phases, but, instead, it enables capacity to be available immediately without additional approvals.
The Recap
- Deepgram raised default concurrency limits across voice products.
- Voice Agent API defaults increased to 45 and 60 connections.
- Customers can upgrade to Growth Plan for higher concurrency.
