livekit /livekit

End-to-end realtime stack for connecting humans and AI

★ 22,222⑂ 2,567👁 176⑃ 4,073● 1,062▤ 90
GoApache-2.0Created 2020-09-30docs.livekit.io
More from this author
Download: v1.13.11v1.13.11 ↓1.13.112026-10-25
+12
Daily
+201
Weekly
22,222
stars
2,567
forks
176
watchers
4,073
prs
1,062
issues
90
releases

README

LiveKit: Realtime infrastructure for voice, video, and AI agents

LiveKit is an open source platform for building voice, video, and physical AI agents.
This repository is the LiveKit server: a scalable, distributed WebRTC SFU that moves realtime audio, video, and
data between people, devices, and AI models. The SDKs, agents frameworks, and companion services are linked in
the table at the bottom of this page.

LiveKit's server is written in Go, using the awesome Pion WebRTC implementation.










 [!IMPORTANT]
 If you're building Voice AI, LiveKit Agents is the SDK for code-first realtime voice agents. STT, LLM, TTS, turn detection, expressive speech, keyterm accuracy, tool usage, and telephony all come bundled in the framework. It's available in both Python and Node.js.

 LiveKit Agents is a great fit for customer support, healthcare, financial services, and outbound calling. Try the live demos.

 python
  agent.py
 from livekit import agents
 from livekit.agents import Agent, AgentServer, AgentSession, STTContextOptions, TurnHandlingOptions, inference

 server = AgentServer()


 @server.rtc_session(agent_name="my-agent")
 async def my_agent(ctx: agents.JobContext):
     session = AgentSession(
         stt=inference.STT(model="deepgram/nova-3", language="multi"),
         llm=inference.LLM(model="google/gemma-4-31b-it"),
         tts=inference.TTS(model="inworld/inworld-tts-2", voice="Ashley"),
         turn_handling=TurnHandlingOptions(turn_detection=inference.TurnDetector()),
         stt_context_options=STTContextOptions(keyterms=["LiveKit", "Acme Corp"]),
         expressive=True,
     )
     await session.start(room=ctx.room, agent=Agent(instructions="You are a helpful voice AI assistant."))
     await session.generate_reply(instructions="Greet the user and offer your assistance.")


 if __name__ == "__main__":
     agents.cli.run_app(server)
 

 Models come from LiveKit Inference with no per-provider API keys, and LiveKit Cloud handles deployment and observability. Visit the docs for more info at docs.livekit.io/agents.

 Using a coding agent? Install the LiveKit skill with npx skills add livekit/agent-skills and add the docs MCP at https://docs.livekit.io/mcp/ (see coding agent support).


 Used in production by

LiveKit carries billions of calls a year for companies including Salesforce, Nvidia, Oracle, SAP,
Deutsche Telekom, Spotify, Tinder, Coursera, Headspace, Skydio, Retell, Decagon, Cresta, and HeyGen. Read how
Assort Health, Playback, and
Polymath Robotics use it, or see more customers.

 Features

-   Scalable, distributed WebRTC SFU (Selective Forwarding Unit)
-   People, devices, and AI agents join the same room as participants, with
    agent dispatch to route agents in automatically or on demand
-   Modern, full-featured SDKs for web, mobile, desktop, embedded, and server
-   Built for production, supports JWT authentication
-   Robust networking and connectivity, UDP/TCP/TURN
-   Easy to deploy: single binary, Docker or K

Top issues

Latest release

v1.13.11 v1.13.11 2026-10-25
Changed

- prometheus: count keyframes received and forwarded (5060)
- config-sample: add rtc.rtx_budget (5059)
- sfu: cap retransmissions per down track (5054)
- sfu: read lock for BufferBase.GetPacket (5055)
- ci: list failed tests after the test step (5057)
- rtc: log the API key that issued the join token (5050)
- rtc: measure subscribe pacing delay in shadow (5046)
- Add StreamType accessor to telemetry StatsKey (5047)
- Extend the TTL of data channel cache to 10s (5045)
- agent: run the agent name lookup off the worker pool (5043)
- Set the psrpc request timeout for PerformRp

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON