LiveKit
by LiveKit
Open-source real-time voice, video and AI agent infrastructure behind ChatGPT's voice mode, with a managed LiveKit Cloud
LiveKit is an open-source platform for real-time audio, video and voice AI agents, combining a WebRTC media server, the LiveKit Agents framework and a managed LiveKit Cloud. It is for engineering teams shipping voice agents, telephony bots and live video apps that need low latency at production scale without building media infrastructure themselves.
LiveKit started in 2021 as an open-source project for apps that send real-time audio and video, founded by Russ d'Sa and David Zhao. Its core is a WebRTC media server built around a Selective Forwarding Unit, which forwards media streams without decoding or re-encoding them, plus client SDKs for web, mobile, Flutter and Unity. The LiveKit Agents framework (Apache-2.0, with a separate Node.js version) lets developers attach Python or Node.js programs to a room as participants, mixing STT, LLM, TTS and realtime API plugins, with a transformer-based semantic turn detection model, native MCP support, job dispatch, a test framework and telephony for placing or receiving phone calls. LiveKit Cloud adds managed agent deployment, observability, inference credits, phone numbers, SIP, recording and a global edge network, priced from a free Build tier to custom Enterprise contracts. The company says its infrastructure powers ChatGPT's Advanced Voice Mode, and TechCrunch and SiliconANGLE list OpenAI, xAI, Salesforce, Tesla, Meta and Spotify among users, along with 911 operators. On January 22, 2026 LiveKit raised a $100 million Series C at a $1 billion valuation led by Index Ventures, bringing total funding to about $183 million per Tracxn. The Agents repository had about 15k GitHub stars when checked on October 30, 2026. Comparisons generally put LiveKit as the integrated, production-ready option against Pipecat's more flexible, assemble-it-yourself framework.
A CTO or head of platform engineering shipping voice agents or live video to customers at scale, who wants managed media infrastructure and telephony instead of running WebRTC servers.
Production voice agent infrastructure (transport, telephony, deployment and observability) on one platform, open source if you want to self-host.
At a Glance
- Category
- Infrastructure & Cloud
- Pricing
- Freemium, Usage-based, Subscription
- Target Market
- CTOs, Enterprise Developers, Heads of Platform Engineering, AI Engineers
- Deployment
- Open-source, Cloud-first, Self-hosted
- Founded
- 2021
Key Features
- ✓LiveKit Agents framework
Open-source Python and Node.js framework that joins agents to rooms as participants and mixes STT, LLM, TTS and realtime plugins.
- ✓Semantic turn detection
A transformer model judges when the user has finished speaking, which cuts the interruptions that silence-only detection causes.
- ✓Built-in telephony
Phone numbers, inbound minutes and third-party SIP let the same agent take or place phone calls without a separate carrier integration.
- ✓SFU media server
Forwards audio and video streams without re-encoding them, keeping latency and bandwidth low for large or multi-party sessions.
- ✓Managed agent deployment and observability
LiveKit Cloud hosts agents with instant rollback, session recordings, metrics export and simulations, so teams can debug calls after the fact.
- ✓Enterprise controls
Region pinning, private links, role-based access, SOC 2 Type II reports and HIPAA on Scale, with SSO and SLAs on Enterprise.
Capabilities
Use Cases
- •AI phone receptionist
Answer inbound calls on a LiveKit number, route intents with an LLM, and record sessions for quality review and compliance.
- •Consumer voice assistant
Stream low-latency voice between a mobile app and an LLM, the same pattern LiveKit says powers ChatGPT's Advanced Voice Mode.
- •Human plus agent meetings
Add an AI agent as a participant in a live video room to transcribe, answer questions or take notes for the people on the call.
- •Self-hosted regulated deployment
Run the open-source server and Agents framework on your own infrastructure to keep audio inside a controlled network boundary.
Ideal For
Best For
- ✓Customer-facing voice agents that must handle thousands of concurrent sessions
- ✓Phone agents that need numbers, inbound minutes and SIP trunking on the same platform
- ✓Multi-participant rooms where humans and AI agents share a live audio or video session
- ✓Teams that want a free self-hosted open-source stack with an upgrade path to a managed cloud
- ✓Robotics and computer-vision apps that stream real-time video to models
Not Ideal For
- ✗Teams that need non-linear or parallel processing inside one pipeline, which comparisons say requires event-hook workarounds in LiveKit Agents
- ✗Small teams that want a no-code voice bot; Vapi or Retell AI will get a simple agent live faster
- ✗High-volume buyers who have not modelled per-minute costs, since agent minutes, telephony and bandwidth all meter separately
Integrations
Deployment
Market Analysis
Pros
- ✓Published pricing with a free tier and clear per-minute overages
- ✓Turn detection model praised by Hacker News practitioners
- ✓Proven at very large scale with OpenAI, xAI and Meta as named users
- ✓Self-hosting is a real option; one HN commenter runs a self-hosted fleet serving 10,000+ users
Cons
- ✗Its sequential agent pipeline is opinionated; parallel or non-linear processing needs workarounds (Chanl comparison)
- ✗Per-minute costs add up at volume, with one comparison estimating $4,000 to $6,000 per month at 100K minutes before model bills
- ✗HN users reported provider compatibility gaps (Gemini, Deepgram Voice Agent API) and difficulty measuring per-call latency and cost without extra tooling
- ✗SSO is gated to the custom-priced Enterprise plan
Pricing
Build
$0
- ✓1,000 agent session minutes
- ✓5 concurrent agent sessions
- ✓1 agent deployment
- ✓1 free US local number
- ✓Community support
Ship
From $50/mo
- ✓5,000 agent minutes then $0.01/min
- ✓20 concurrent agent sessions
- ✓Custom voices
- ✓Instant rollback
- ✓Email support
Scale
From $500/mo
- ✓50,000 agent minutes then $0.01/min
- ✓Region pinning and private links
- ✓SOC 2 Type II reports
- ✓HIPAA
- ✓Inference discounts
Enterprise
Contact for pricing
- ✓Volume pricing including inference
- ✓SSO
- ✓Support SLA
- ✓Shared Slack and solutions engineer
LiveKit Cloud publishes list prices: Build is free, Ship starts at $50/month and Scale at $500/month, each with included agent minutes then $0.01 per agent minute. Telephony, SIP, WebRTC minutes, bandwidth, recordings and inference meter separately. HIPAA and region pinning start at Scale; SSO is Enterprise only. The open-source server and Agents framework are free to self-host. Prices read from livekit.com on October 30, 2026.
Security & Compliance
Connect
Sources
This page was written from 7 sources, 6 on domains other than livekit.com.
- 1.livekit.com — pricingvendor
- 2.github.com — agents
- 3.techcrunch.com — voice ai engine and openai partner livekit hits 1b valuation
- 4.siliconangle.com — livekit raises 100m 1b valuation scale real time ai media pl
- 5.channel.tel — pipecat vs livekit voice framework decision
- 6.cekura.ai — livekit alternatives
- 7.hn.algolia.com — hn.algolia.com
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Akamai Inference
Managed open-weight model inference on Akamai Cloud with an OpenAI-compatible API, serverless or dedicated GPUs
Grafana Assistant
AI observability agent in Grafana Cloud that writes PromQL, LogQL and TraceQL from plain English, investigates incidents and proposes code fixes as pull requests
Inferact Platform
Managed vLLM inference from the creators of vLLM, on your own GPUs or on Inferact Cloud
Ollama
Run open-weight LLMs locally with one command, then burst to Ollama's cloud on the same API