Telnyx launched “LiveKit on Telnyx,” a hosted platform that runs LiveKit voice AI agents on Telnyx-owned GPU and telephony infrastructure, the company announced on April 6. The platform promises 50% lower speech-to-text and text-to-speech costs compared to LiveKit Cloud, sub-200ms round-trip latency, and no session fees during the beta period, according to the press release.
How It Works
Developers package their existing LiveKit agent with a Dockerfile and deploy via API. No code changes required. Telnyx handles infrastructure provisioning, GPU allocation, and telephony connectivity. Audio stays on the Telnyx network end-to-end, which eliminates the variable latency that comes from routing through external speech APIs, according to the press release.
The cost advantage comes from vertical integration. Most voice AI platforms resell third-party APIs for speech processing and telephony, stacking margins at each layer. Telnyx owns the carrier network, GPU clusters, and telephony stack, cutting out intermediaries. LiveKit Cloud currently charges $0.01 per minute per active session; Telnyx is waiving that fee during beta.
Enterprise Telephony Built In
Unlike standalone LiveKit deployments, LiveKit on Telnyx includes carrier-grade SIP capabilities: AMR-WB codec support, call recording, call transfers, custom trunk configurations, and verified identity. These are native to the platform rather than integrated from third parties, according to the press release.
“Voice AI is moving from prototype to production,” said David Casem, CEO and co-founder of Telnyx. “That transition demands infrastructure built for enterprise reliability and compliance.”
The Voice Agent Infrastructure Layer
LiveKit has become the dominant open-source framework for building voice AI agents. Telnyx is positioning itself as the managed deployment layer on top of it, similar to how cloud providers offer managed Kubernetes rather than asking developers to run their own clusters. For teams building customer service agents, IVR replacements, or real-time voice assistants, the pitch is straightforward: keep your existing LiveKit code, deploy it on cheaper and faster infrastructure, and get enterprise telephony without additional integrations.