Both are open source under the Apache 2.0 license, both are built on WebRTC, and both route video through a selective forwarding unit (SFU) instead of mixing it on the server. The real difference is how much is finished when you start: Jitsi hands you a complete meeting app, LiveKit hands you the building blocks.
Quick answer: Jitsi or LiveKit?
Choose Jitsi Meet for a private or embedded meeting app with an existing interface. Choose LiveKit when your team needs control over the real-time experience or is building an AI agent. LiveKit Meet is a real open source conferencing app, so this is not a choice between having a UI and having none. The decision is how much of that UI and its operating model fit your requirements. [S1][S3][S5]
This comparison is for developers and administrators choosing a platform. If you just want to enter a meeting, use our join a Jitsi meeting guide.
At a glance
| Jitsi Meet | LiveKit | |
|---|---|---|
| What you get | A complete meeting app | A media server plus client SDKs |
| User interface | Included: lobby, moderation, chat, polls, breakout rooms, whiteboard | Adapt the LiveKit Meet app or build with SDKs |
| Development work | Configure and embed an existing meeting interface | Adapt LiveKit Meet or build an interface with SDKs |
| Embedding | IFrame API and JWT | Native SDKs: web, iOS, Android, Flutter, React Native, Unity and more |
| Recording | Optional Jibri service, configured separately | Optional Egress service, configured separately |
| Transcription | Optional Jigasi transcription and a configured speech service | Through Agents and a speech-to-text provider, configured separately |
| AI voice and video agents | Not a focus | A core use case: Agents framework, telephony, inference |
| Phone dial-in | Jigasi (SIP gateway) | LiveKit SIP |
| Server components | Prosody, Jicofo, Jitsi Videobridge (Java), nginx | LiveKit server (Go), Redis for multi-node, optional Egress, Ingress, SIP |
| Hosted option | 8x8 JaaS, per monthly active user | LiveKit Cloud, per minute and per GB beyond plan allowances |
| License | Apache 2.0 | Apache 2.0 |
Choose Jitsi when
- You need an existing meeting interface. Teams, schools, clinics and communities get a proven interface with lobby, moderator controls, screen sharing, breakout rooms and polls when enabled in your deployment.
- You want to embed meetings with little frontend work. The IFrame API drops a full meeting into your page, and JWT authentication lets your backend control access.
- Data has to stay on your servers. Recording and transcription can run entirely inside your own infrastructure.
- You want control over infrastructure costs. Budget for compute, bandwidth, recording storage, monitoring and maintenance. A fixed server bill does not imply unlimited capacity or fixed total cost. See what self-hosting costs.
Choose LiveKit when
- The real-time experience is your product: live shopping, interactive streaming, telehealth inside your own app, multiplayer, or classroom tools with a custom layout.
- You are building AI voice or video agents. LiveKit’s Agents framework, SIP telephony and inference integrations provide a documented path for these applications.
- You need native mobile and game-engine SDKs with fine control over tracks, layouts and data channels.
- You have frontend and mobile developers to build and maintain the interface, to adapt the example meeting app or build your own product.
Self-hosting compared
| Jitsi Meet | LiveKit | |
|---|---|---|
| Install | Debian packages or docker-jitsi-meet | Generator for Docker Compose and Caddy, or Kubernetes |
| Public ports | 443/tcp, 10000/udp, 80/tcp for certificates | 7880 behind a TLS load balancer, 7881/tcp, 50000-60000/udp |
| TURN for strict networks | Separate coturn on 3478/udp and 5349/tcp | Built-in TURN on 3478/udp and 5349/tcp (or 443) |
| Scaling | Add videobridges behind one signalling server | Add nodes that share state through Redis |
Do not choose solely by counting ports. LiveKit supports UDP multiplexing on 7882 instead of the default media range. Jitsi also needs certificate renewal and configured TURN fallback ports; Docker and Debian package deployments differ. [S2][S4] Our Jitsi ports and firewall reference and server requirements cover the Jitsi side in detail.
Hosted pricing compared
If you do not want to run servers, both have a hosted service. Prices checked on 2026-10-13. Amounts below are USD and can change:
- LiveKit Cloud: a free Build plan with 5,000 WebRTC minutes and 50 GB of downstream data a month; Ship from $50 a month with 150,000 minutes and 250 GB included, then $0.0005 per minute and $0.12 per GB; Scale from $500 a month. AI agent sessions, telephony and inference have separate allowances and charges. [S6]
- 8x8 JaaS: billed per monthly active user, with extra per-minute fees for recording and transcription. See JaaS vs self-hosted Jitsi for the numbers and when self-hosting is cheaper.
The cheaper model depends on your usage. Frequent meetings increase participant-minutes, while user-based pricing depends on how many distinct people use the service.
What does a meeting minute cost?
LiveKit bills WebRTC time per participant, not once per room. For example, 10 participants connected for 60 minutes use 600 participant-minutes. On Ship, if the included allowance is already exhausted, those minutes cost $0.30 at $0.0005 per minute. This is an illustration, not a complete bill: add the plan fee, downstream data and any recording, telephony or agent usage. Current transport billing uses seconds with a ten-second minimum per WebRTC session. [S6][S7]
The free Build allowances are hard caps; new requests fail after an allowance is exhausted. Check quotas before relying on it for production. [S8]
JaaS uses monthly active users rather than room minutes. Self-hosting removes the hosted participant-minute charge but still incurs infrastructure and operating costs. Compare all three with the same participant count, duration, video quality, recording retention and support requirements.
Before choosing: test the same workflow
- Test your intended guest access, organizer permissions and branding on both platforms.
- Run a three-person call from different networks, including a network that blocks UDP. Check audio, video, screen sharing and reconnect behavior.
- If you need recording or captions, configure the required services and verify the actual output. A recording button alone is not evidence of a working recording pipeline.
- Estimate monthly usage using participant-minutes, downstream data, storage and operating effort. Include separate AI and telephony charges when relevant.
- Compare the frontend work your team actually needs. Avoid assuming either platform is faster or cheaper without this trial.
These are proposed evaluation steps, not a claim that we ran a comparative benchmark.
Keep a decision record
Download the blank platform evaluation worksheet and run your actual access, media and operating workflow on each platform. Record release/version, workload and observed results beside each requirement. A blank or untested row means unknown, not a pass. Keep a failed requirement visible even if the other platform has more features.
For a deployment budget, use the server requirements guide, sizing calculator and cost worksheet. A comparison table cannot replace checking your integration and operating requirements.
Can you switch later?
Not without rework. Room creation, access tokens and embedded interfaces use different APIs, and a LiveKit app is built on its own SDK and server APIs. Both use JWT-based authentication, but their token claims and permission models are not interchangeable. If you start with Jitsi to get meetings running now and later build a custom product, plan the LiveKit app as a new frontend rather than a migration. Both can run side by side while you move users over.
Our view
If the goal is “our people need private, branded video meetings”, Jitsi provides an existing meeting interface, and our one-click hosting removes the setup. If the goal is “we are building a new real-time product or an AI agent”, LiveKit deserves a serious evaluation, with its documented SDK and Agents workflows. We have not benchmarked development time, latency or capacity head to head.
Sources
Original technical review: October 13, 2026. LiveKit’s published plan fees, WebRTC minutes and downstream allowances rechecked October 20, 2026; the evaluation worksheet was added on that date. Recheck pricing and deployment requirements before purchasing or opening firewall ports.
- [S1] Jitsi Meet source and license
- [S2] Jitsi Debian/Ubuntu installation and firewall
- [S3] LiveKit Meet app and setup
- [S4] LiveKit ports and firewall
- [S5] LiveKit Agents introduction
- [S6] LiveKit Cloud pricing
- [S7] LiveKit billing units and minimums
- [S8] LiveKit quotas and free-plan caps
Further documentation
- LiveKit documentation: ports and firewall
- LiveKit documentation: deploy to a VM
- LiveKit pricing (checked 2026-10-13)
- LiveKit server on GitHub (Go, Apache 2.0) and LiveKit Meet example app
- Jitsi Meet on GitHub (Apache 2.0)
- Jitsi handbook: self-hosting and firewall