News · 2026-09-29
Deepgram ships Flux TTS: conversation-native speech with 80ms first audio for voice agents
Deepgram's Flux TTS reached general availability on the /v2/speak endpoint (launched 12 August 2026): a text-to-speech model built for live voice agents that holds the whole conversation in memory, handles interruptions natively, starts responding in as low as 80 milliseconds, and ships 36 English voices across seven accents. Deepgram passed $100 million in annual recurring revenue at launch.
What it means for buyers
Voice-agent stacks are consolidating around paired speech models: Flux TTS is the default speak provider in Deepgram's Voice Agent API, runs beside Flux STT behind one API, and plugs into the platforms agent teams already use, including Pipecat, LiveKit, jambonz, and Twilio. Buyers comparing voice-agent platforms should ask which speech stack sits underneath, because cross-turn consistency and interruption handling are now model properties, not integration projects. Deepgram names Vapi and Granola among its customers, so the platforms in our voice-agent and meeting-notes comparisons are directly riding this infrastructure; the free window (up to 45 concurrent connections, 5 in EU/AU) closed 12 September and standard pricing now applies.
Where to check it
Sources
- Deepgram: Introducing Flux TTS — launch post, checked 2026-09-30
- Deepgram press release: Text-to-Speech Comes of Age — GA terms, $100M ARR, customer list, 12 August 2026
- Deepgram: Flux TTS at general availability — endpoint, voices, and pricing-transition details