Skip to content
SPECIFICTION
Get listed

LiveKit Agents

livekit.com

From $50/mo Free plan Open source WebWindowsmacOSLinuxiOSAndroid
Visit livekit.com
Spec score 9.1 #1 of 32 in AI Voice Agents

What it is

LiveKit Agents is an open source framework and cloud platform for building voice, video, and physical AI agents. Python or Node.js programs can join LiveKit rooms as realtime participants, and deployments can run on LiveKit Cloud or a developer-chosen environment. Agents can process realtime voice, video, and text input and produce output in any modality. The framework supports LLM tool use, frontend RPC calls, multi-agent handoffs, and connections to phone systems through native SIP. LiveKit Inference provides models from providers including OpenAI, Google, AssemblyAI, Deepgram, Cartesia, and ElevenLabs; open source plugins connect to other providers. LiveKit Cloud includes agent insights such as session replay, transcripts, trace spans, logs, and reports. The free Build plan includes 1,000 agent session minutes, five concurrent sessions, and one deployment. Ship starts at $50 a month and includes 5,000 minutes, then $0.01. LiveKit says prompts, audio, and completions handled by LiveKit Inference and its model providers are not logged or retained. It lists SOC 2 Type II, GDPR, HIPAA, CCPA / CPRA, and the EU–US Data Privacy Framework; PCI DSS and ISO 27001 are in progress.

Spec sheet

What it does
LiveKit Agents is a realtime framework and cloud platform for building voice, video, and physical AI agents.Read 15 Oct 2026 · livekit.com
Product
LiveKit Agents is an open source framework and cloud platform for building voice, video, and physical AI agents.Read 29 Sep 2026 · livekit.com
Deployment
Agents can be deployed to LiveKit Cloud or to a custom environment of the developer’s choice.Read 29 Sep 2026 · docs.livekit.io
Integrations
LiveKit Inference provides models from providers including OpenAI, Google, AssemblyAI, Deepgram, Cartesia, and ElevenLabs, while open source plugins connect to additional providers.Read 29 Sep 2026 · docs.livekit.io
Security
LiveKit says prompts, audio, and completions handled by LiveKit Inference and its model providers are not logged or retained.Read 29 Sep 2026 · livekit.com
Compliance
LiveKit lists SOC 2 Type II, GDPR, HIPAA, CCPA / CPRA, and the EU–US Data Privacy Framework; it says PCI DSS and ISO 27001 are in progress.Read 29 Sep 2026 · livekit.com
Support
Build includes community support, Ship includes email support, and Enterprise includes a support SLA.Read 29 Sep 2026 · livekit.com
Headquarters
San Francisco, California, United StatesRead 23 Sep 2026 · livekit.com
Agent Builder
Agent Builder is a browser-based, no-code tool included with LiveKit Cloud and is unavailable in self-hosted deployments.Read 15 Oct 2026 · docs.livekit.io
Agent orchestration
The documentation covers agent sessions, workflows, handoffs, tasks, task groups, tools, and external-data retrieval.Read 15 Oct 2026 · docs.livekit.io
Client SDKs
LiveKit offers frontend and client SDKs for web, mobile, and other platforms, while its Unity SDK supports macOS, Windows, Linux, iOS, and Android.Read 15 Oct 2026 · linkedin.com
Company founded and size
LinkedIn lists LiveKit as founded in 2021 with 51–200 employees.Read 15 Oct 2026 · au.linkedin.com
Company headquarters
LiveKit’s official site lists 369 Sutter Street in San Francisco, California.Read 15 Oct 2026 · livekit.com
Company legal name
Crunchbase identifies the legal entity as LiveKit Incorporated.Read 15 Oct 2026 · crunchbase.com
Company location
LiveKit lists its address as 369 Sutter St, San Francisco, California.Read 29 Sep 2026 · livekit.com
Documentation
Official documentation includes quickstarts, recipes, SDK references, API references, deployment guidance, telephony, models, testing, and troubleshooting material.Read 15 Oct 2026 · docs.livekit.io
Encryption
WebRTC media uses SRTP with DTLS-SRTP exchange, SIP supports TLS and SRTP, and stored data uses AES-256 encryption.Read 15 Oct 2026 · livekit.com
HIPAA
LiveKit says BAAs are available for Scale and Enterprise customers handling protected health information.Read 29 Sep 2026 · livekit.com
Inference retention
LiveKit says prompts, audio, and model outputs sent through LiveKit Inference are not logged or retained by LiveKit or its model providers.Read 15 Oct 2026 · livekit.com
Licence
The Agents framework is Apache-2.0 licensed, while LiveKit turn-detection models use a separate LiveKit Model License.Read 15 Oct 2026 · github.com
Limit: cloud-only capabilities
CloudTalk’s comparison says Agent Builder, LiveKit Inference, adaptive interruption handling, Krisp noise cancellation, and built-in observability are unavailable when self-hosting.Read 15 Oct 2026 · cloudtalk.io
Limit: operational complexity
Airframe’s review notes that self-hosting can require command-line and server-management skills and that production deployments carry operational complexity.Read 15 Oct 2026 · airframe.ai
Limit: pricing concerns
A G2 review praised capability but specifically identified pricing as a concern.Read 15 Oct 2026 · g2.com
Media and modalities
Agents can process realtime voice, video, and text input and produce output in any modality.Read 29 Sep 2026 · docs.livekit.io
Model integrations
LiveKit Inference and plugins connect agents to OpenAI, Google, AssemblyAI, Deepgram, Cartesia, ElevenLabs, and other providers.Read 15 Oct 2026 · docs.livekit.io
Multimodal agents
The framework supports realtime spoken interaction, video input and output, and physical-agent applications.Read 15 Oct 2026 · docs.livekit.io
Node requirement
The Node.js quickstart requires Node.js 20 or newer and pnpm 10.15.0 or newer.Read 15 Oct 2026 · docs.livekit.io
Notable limit
Agent Builder is intended for simple voice agent prototypes and produces Python code for deployment to LiveKit Cloud.Read 29 Sep 2026 · docs.livekit.io
Observability
LiveKit Cloud includes transcripts, traces, session recordings, and logs for agent sessions, while self-hosted deployments can use OpenTelemetry hooks.Read 15 Oct 2026 · docs.livekit.io
OpenAI integration
The OpenAI integration supports GPT-Live, the Realtime API, GPT-4.1, GPT-4o, GPT-5, OpenAI speech recognition, and OpenAI text-to-speech.Read 15 Oct 2026 · docs.livekit.io
OpenAI-compatible APIs
The OpenAI plugin can connect to compatible endpoints such as Azure, DeepSeek, Fireworks, Groq, OpenRouter, Perplexity, Together, and xAI.Read 15 Oct 2026 · docs.livekit.io
Praise: ease and cost
A Trustpilot reviewer said LiveKit was smooth for voice-calling agents, reasonably priced per minute, and supported by helpful customer service.Read 15 Oct 2026 · trustpilot.com
Praise: integration and scalability
A G2 reviewer praised easy integration, scalability, voice-agent capabilities, noise suppression, echo cancellation, model-provider choice, and SIP telephony.Read 15 Oct 2026 · g2.com
Praise: self-hosting and community
Another Trustpilot reviewer praised the ability to deploy with a few containers, the open Go-based server components, community support, and perceived stability.Read 15 Oct 2026 · trustpilot.com
Programming languages
Agents can be developed with Python or Node.js, with first-class support for speech-to-text, language-model, and text-to-speech pipelines.Read 15 Oct 2026 · livekit.com
Python requirement
The Agents quickstart requires Python 3.10 or newer and uses uv as its package manager.Read 15 Oct 2026 · docs.livekit.io
Rating on G2
4.8 out of 5 from 2 reviewsRead 15 Oct 2026 · g2.com
Rating on Trustpilot
3.8 out of 5 from 2 reviewsRead 15 Oct 2026 · trustpilot.com
SDKs
The framework lets Python or Node.js programs join LiveKit rooms as realtime participants.Read 29 Sep 2026 · docs.livekit.io
Security compliance
LiveKit states that its cloud service has SOC 2 Type II, GDPR, CCPA/CPRA, EU-US Data Privacy Framework certification, and HIPAA BAAs for Scale and Enterprise customers.Read 15 Oct 2026 · livekit.com
Self-hosted limitations
Self-hosting requires the customer to operate and scale agent servers and supply model-provider keys instead of using managed hosting and built-in inference.Read 15 Oct 2026 · docs.livekit.io
Self-hosting
Organizations can run LiveKit servers and Agents on local machines, virtual machines, Kubernetes, or distributed multi-region infrastructure.Read 15 Oct 2026 · docs.livekit.io
Series A funding
LiveKit announced a $22.5 million Series A on June 4, 2024, led by Altimeter with Redpoint participation.Read 15 Oct 2026 · livekit.com
Series C news
LiveKit announced a $100 million Series C on January 22, 2026, led by Index Ventures, at a reported $1 billion valuation.Read 15 Oct 2026 · livekit.com
Telephony
LiveKit provides SIP-based inbound and outbound calling and can bridge phone calls into LiveKit rooms.Read 15 Oct 2026 · docs.livekit.io
Testing
The platform provides text simulations, audio simulations, session recordings, and turn-by-turn observability events.Read 15 Oct 2026 · livekit.com
Tools and workflows
Agents support LLM tool use, frontend RPC calls, and multi-agent handoffs.Read 29 Sep 2026 · docs.livekit.io
Turn taking
Built-in turn detection identifies when a speaker has finished and supports interruption handling during responses.Read 15 Oct 2026 · livekit.com

Plans and prices

Build Free $0/mo 1,000 agent minutes · 5 concurrent sessions · 1 deployment · 100,000 observability entries · 1 free US number Read 15 Oct 2026 · livekit.com
Ship $50 a month $50/mo 5,000 agent minutes · 20 concurrent sessions · 2 deployments · 500,000 observability entries · custom voices Read 15 Oct 2026 · livekit.com
Scale $500 a month $500/mo 50,000 agent minutes · up to 600 concurrent sessions · 4 deployments · 5,000,000 observability entries · HIPAA/security reports Read 15 Oct 2026 · livekit.com
Enterprise Priced on request Custom Volume pricing · shared Slack channel · SSO · support SLA Read 15 Oct 2026 · livekit.com

Every LiveKit Agents plan and price →

Strong points and limits

Strong points

  • Agents can process voice, video, and text in realtime.
  • Deploy to LiveKit Cloud or a chosen environment.
  • Native SIP connects agents to phone systems.
  • Build includes 1,000 session minutes and five concurrent sessions.
  • Cloud insights include session replay, transcripts, logs, and reports.

Limits

  • Build allows five concurrent agent sessions.
  • Ship starts at $50 a month.
  • Ship includes 5,000 minutes, then $0.01.
  • Agent Builder is intended for simple voice prototypes.

Who it suits

LiveKit Agents suits developers building realtime voice, video, text, or physical AI agents in Python or Node.js. Teams can choose LiveKit Cloud or a custom deployment environment, with SIP support for phone-system connections.

LiveKit Agents in depth

Read the full write-up

Overview

LiveKit Agents is a developer framework and cloud platform for building realtime voice, video, and physical AI agents. It suits teams that want to write agents in Python or Node.js and choose between managed deployment and their own environment. Its range of media and telephony capabilities is a strength; the trade-off is that this is a framework to build with, not a finished assistant to configure.

Key features

Agents join LiveKit rooms as realtime participants, taking voice, video, and text input and producing output in any modality. Tool use, frontend RPC calls, and multi-agent handoffs make it possible to build workflows beyond a single conversational exchange, but require a developer-led approach.

Deployment can run on LiveKit Cloud or in a custom environment. LiveKit Inference connects to models from OpenAI, Google, AssemblyAI, Deepgram, Cartesia, and ElevenLabs, with open-source plugins extending provider choices. Native SIP supports phone-system connections; inbound and outbound calling, custom phone numbers, and human handoff suit teams building phone-based agents.

LiveKit Cloud’s agent insights include session replay, transcripts, trace spans, logs, and session reports. LiveKit says prompts, audio, and completions handled by LiveKit Inference and its model providers are not logged or retained. It lists SOC 2 Type II, GDPR, HIPAA, CCPA / CPRA, and the EU–US Data Privacy Framework; PCI DSS and ISO 27001 are in progress. BAAs are available to Scale and Enterprise customers handling protected health information. Those assurances and options matter for sensitive deployments, though the BAA terms place a practical threshold at the higher-priced tiers.

Agent Builder is aimed at simple voice-agent prototypes and produces Python code for deployment to LiveKit Cloud. It is useful for getting a basic prototype started, but not a substitute for the framework when a project needs a different deployment target or more involved design.

Pricing

The free Build plan costs $0/mo and includes 1,000 agent session minutes, five concurrent sessions, one deployment, $2.50 in inference credits, and community support. It is a credible starting point, but the session and deployment caps make it better suited to prototyping than sustained workloads.

Ship starts at $50/mo and raises the allowance to 5,000 session minutes, 20 concurrent sessions, and two deployments, with email support. Usage above the included minutes costs $0.01 per minute. This is the sensible paid entry point for a small production workload, though its concurrency and included minutes still set clear ceilings.

Scale starts at $500/mo, including 50,000 session minutes, up to 600 concurrent sessions, four deployments, region pinning, security reports, and HIPAA support. Additional minutes cost $0.01 each. It fits higher-volume or regulated deployments; the jump in monthly cost is difficult to justify for smaller workloads.

Enterprise has custom pricing, with volume pricing, a shared Slack channel, SSO, and a support SLA. It is aimed at organizations that need those commercial and support arrangements. Build includes community support, Ship email support, and Enterprise a support SLA; Scale’s stated package emphasizes security and capacity instead.

Platforms

LiveKit Agents supports Android, iOS, web, Windows, macOS, Linux, self-hosted environments, and an API. That breadth helps teams target both app and server deployments, while the choice of runtime and environment still calls for development work.

Who it's for

Choose LiveKit Agents if you have developers building realtime agents that may span app, web, or phone interactions and want control over deployment, model connections, and workflow design. It is less suitable if you need a ready-made voice assistant or a prototype without a path into Python development.

Pros and cons

Pros: Realtime voice, video, and text plus multimodal output support more than voice-only use cases.

Pros: Cloud or custom deployment, model integrations, plugins, and native SIP give developers room to shape the stack.

Pros: Session replay, transcripts, traces, logs, and reports offer several ways to inspect agent activity.

Cons: The free plan’s 1,000 minutes, five concurrent sessions, and single deployment are restrictive beyond early development.

Cons: Ship’s overage charge and Scale’s $500/mo starting price make sustained or higher-volume use a budget decision, not just a framework choice.

Cons: Agent Builder is limited to simple voice prototypes and deploys its generated Python code to LiveKit Cloud.

Alternatives

Consider Rapida AI if a free one-agent starting plan with bring-your-own STT, TTS, LLM, and SIP is preferable to LiveKit’s integrated inference options.

Intervo is another option to compare for its self-hosted open-source plan, which includes full source code under an MIT License.

Tone may suit teams seeking a self-hosted plan with unlimited agents, calls, and workflows, while supplying their own LLM, voice, and telephony.

Yellow.ai is worth considering when a free plan centered on web and mobile platforms is the priority.

Ring AI offers a free tier with 100 minutes per month for testing, a narrower starting point for phone-agent evaluation.

Vapi is a paid alternative with usage-based billing and $5 in free credits to get started.

Dialzara is a paid alternative with a Business Lite plan at $29 a month, including 60 minutes and $0.48 per minute in overages.

My AI Front Desk is an alternative with a Business-in-a-Box plan priced at $99 a month, or $79/mo billed annually.

Browse AI Voice Agents or Conversational AI Platforms to compare more options.

Verdict

LiveKit Agents is the stronger fit for developer teams building realtime agents that need multimodal input, telephony, and deployment flexibility. Its free plan lowers the cost of starting, and paid tiers add meaningful capacity and support, but minute and concurrency limits—and the $500/mo Scale starting price—make it a less compelling choice for small workloads or anyone seeking a finished assistant rather than a framework.

Verdict

LiveKit Agents combines an open framework with deployment, inference, telephony, and observability options. Build provides a free starting plan, while higher session use and support are described across paid tiers.

Where Specifiction places it

Alternatives to LiveKit AgentsAll alternatives →