Checklist for Regional Deployments

Deploying LiveKit in a specific geographic region requires careful planning. You must evaluate data residency, traffic routing, infrastructure placement, regulatory compliance, and operational visibility.

This guide provides a structured checklist to help you plan a regional deployment. It is intended as a starting point. Always consult the linked documentation for the most current and authoritative guidance.

The LiveKit docs also contain authoritative documentation for region pinning which supersedes this article in the event of contradiction.

LiveKit Region Pinning

Region pinning controls which LiveKit media servers are selected for rooms and sessions.

Use region pinning when participant media (text, audio, video) must remain within a specific country or economic area. This is commonly required for deployments subject to regulations such as GDPR, HIPAA, or financial compliance frameworks.

Be aware that participants located outside the pinned region may experience increased latency, since they are prevented from connecting to a closer LiveKit region.

More information, including the list of available regions: Region Pinning docs

How to apply: Region pinning is available on the Scale plan or higher. Contact our support team if you need to pin traffic to a specific region. Our team can also help you evaluate redundancy and failover considerations.

Note: Region pinning applies to media servers. It does not automatically cover logs, control-plane metadata, or third-party integrations. Those topics are addressed later in this guide.

Agent Hosting

Agent hosting determines where your AI agents execute and process data.

This is important when agents handle sensitive or regulated information, including prompts or transcripts that must remain within a specific jurisdiction.

LiveKit supports two agent hosting approaches:

• LiveKit Cloud hosting - Run your agent within the same infrastructure that powers LiveKit Cloud. See the Agent deployment docs for a list of supported regions, which we are continually expanding.
• Self-hosting – Deploy and operate your agents in your own infrastructure. This is appropriate when you need a region that is not yet supported by LiveKit Cloud. See the self-hosted deployments docs for details.

If your required region is supported, LiveKit Cloud hosting is generally recommended for operational simplicity.

Model Hosting

Voice AI applications rely on multiple models, including STT, TTS, and LLM components. Some deployments use individual pipeline models, while others use real-time multimodal models. You may also optionally include an avatar.

If you have regional regulatory requirements, you must understand where each model processes data. Even if LiveKit media is pinned to a specific region, external model APIs may process requests elsewhere. Ensure your agent hosting and model endpoints are aligned to avoid unintended cross-region data flows.

This section does not cover each model in detail, but your region-restricted options include:

LiveKit Inference

Currently, U.S. customers can access U.S.-hosted models via LiveKit Inference. See our list of Inference sub-processors for full details. Customers deploying HIPAA-compliant workloads should also review our list of HIPAA eligible products, which includes supported STT, TTS, and LLM services.

We also offer regional Deepgram hosting in Mumbai for STT. See our docs for more information.

What about outside of the US? LiveKit Inference is continually evolving to meet the needs of customers with strict data residency requirements. As of February 2026, we are still expanding global coverage. Refer to the LiveKit inference docs for the most current status.

Plugins

Additional regional options are available through third-party providers via LiveKit plugins:

• Major LLM providers such as OpenAI and Google Gemini offer “inference residency” to restrict processing within defined regions.
• Many enterprise STT and TTS providers support regional hosting.
• Azure OpenAI enables regional deployments within specific Azure regions across LLM, STT, and TTS. Similarly, AWS provides regional hosting across the full AI pipeline.
• Some countries offer locally developed models tailored to national requirements. If your LLM exposes an OpenAI-compatible endpoint, you can use the OpenAI plugin and specify a custom base_url.
• Region-locked self-hosted deployments are possible through providers such as Baseten. You can also run models locally using services such as Ollama.

This is not an exhaustive list. For reliability, consider using multiple providers in combination with a fallback adapter.

SIP Pinning

SIP pinning controls where PSTN ingress and egress occur with respect to LiveKit.

If your telephony traffic must remain in a specific country, for example, to comply with local telecom regulations, you must pin your SIP traffic accordingly.

SIP pinning ensures that calls pass through the LiveKit SIP gateway in the designated region, so be sure to select a carrier who supports that region and consult their documentation for any additional requirements.

More information, including supported regions can be found in our telephony region pinning docs. Also see our documentation for region-specific outbound calls.

How to apply:

• Inbound calls – Use the region-specific SIP endpoint.
• Outbound calls – Specify the region in the trunk configuration.

LiveKit phone numbers are not available in all regions. If the required region is unsupported, you must integrate with an external telephony provider.

Agent Observability

Agent Observability stores and processes agent logs, traces, and audio data in LiveKit Cloud. This data may be considered personal data in regulated environments.

As of the time of writing (February 2026), Observability data is processed and stored in the United States. Customers with strict residency requirements should disable agent observability at the project level.

We are actively working to enable region-specific Observability options. Refer to the Observability docs for the latest updates.

Additional Links

The following resources may be helpful when planning a regional deployment:

• The LiveKit trust center includes compliance documentation such as our SOC2 report, BAA agreement, GDPR and HIPAA materials, pen test results, and related resources.
• Our legal pages include the list of sub-processors, the data processing addendum, and the list of HIPAA eligible services.

Static IP Address Support

Some enterprise customers require static IP addresses for firewall allowlisting.

The list of Static IPs is available in the firewall documentation.

Static IP addresses apply only to specific regions and services and if your region is not listed, you must rely on wildcard domains. If a service is not listed in the documentation, it is not guaranteed to receive an IP address within the static block.

When you pin to a region that supports static IPs, traffic will originate from the documented IP block for that region.

Additional Data Considerations

• Please refer to our DPA for additional information; specific details about region pinning can be found in section 2.5.
• You are responsible for ensuring that any exported data complies with your residency requirements. This includes (but is not limited to) agent logs, session telemetry, or egress.

Worked Examples

Below are example deployment strategies for common regulatory environments.

Any other questions?

If you have additional questions, please contact us via our contact sales link or join our developer community