Telnyx Launches LiveKit on Telnyx to Power Low-Cost, Ultra-Low Latency Voice AI Agents

Telnyx, the renowned Voice AI infrastructure provider, has made public the availability of its newest creation, known as LiveKit on Telnyx, which will allow developers to deploy their voice AI agents for reduced costs, low latency, and enterprise telephony in one package.

The developers do not have to make any changes to the LiveKit agent they had deployed because the LiveKit is part of the infrastructure stack.

Using LiveKit on Telnyx, users are able to bundle their agents into Dockerfiles, deploy via API, and then be up and running on Telnyx-hosted infrastructure, all without setting up any servers or working with any third parties.

This is expected to make it easier for enterprises to adopt production-grade voice AI applications.

Infrastructure Ownership Reshapes the Cost Model

A major differentiator of the new offering is Telnyx’s ownership of the full infrastructure stack.

Unlike many voice AI platforms that depend on third-party APIs for speech processing and telephony – often adding multiple layers of margin – Telnyx owns the carrier network, GPU clusters, and telephony layer.

This infrastructure-first approach significantly improves the economics of voice AI deployments.

According to the company, customers can achieve up to 50% lower speech-to-text (STT) and text-to-speech (TTS) costs compared with LiveKit Cloud on equivalent models.

In addition, Telnyx is waiving session fees during the beta period, eliminating the current $0.01 per minute active session fee that some LiveKit users typically incur.

Also Read: Circana Launches Liquid Activation™ to Build Purchase-Based Audiences in Minutes

Ultra-Low Latency Through Colocated Inference

Performance is another major focus of the launch.

Telnyx hosts STT and TTS models on its own GPU infrastructure, colocated with global telephony points of presence.

Since the audio will always stay inside the Telnyx network itself, there will be no latency created due to any third-party APIs.

Such an architecture allows achieving sub-200 milliseconds of latency in return traffic, making possible conversations much more natural.

The compute layer will also be designed for scalability and enterprise security.

“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. LiveKit on Telnyx gives developers the framework they know with the carrier-grade foundation that enterprise production deployments require.”

Carrier-Grade Telephony Built In

For enterprise voice AI use cases, telephony capabilities are often as critical as AI performance.

LiveKit on Telnyx includes native support for advanced SIP capabilities such as:

• AMR-WB codec support
• call recording
• call transfers
• custom trunk configurations

Unlike many competing solutions, these features are not dependent on external integrations.

They are built directly into the Telnyx platform, improving reliability and simplifying enterprise deployment.

“LiveKit created an excellent agent framework,” said Jon Scalet, Technical Product Manager at Telnyx. “LiveKit on Telnyx extends it with the telephony infrastructure that enterprise deployments require, carrier-grade SIP, verified identity, and compliance built into every call.”

For developers and enterprise teams building production-grade voice AI applications, the launch positions Telnyx as a strong infrastructure partner focused on cost efficiency, latency optimization, and carrier-grade reliability.