LiveKit · Pricing · Checked Nov 1, 2026
LiveKit Cloud pricing: plans and cost per minute
LiveKit Cloud charges $0.01/min for agent session minutes after the included allotment on Ship and Scale. Build is $0/month, Ship starts at $50/month, and Scale starts at $500/month. Session minutes, recording, WebRTC, and SIP have been billed per second, with a ten-second minimum, since August 24, 2026. Checked Nov 1, 2026.
LiveKit Cloud pricing at a glance
| Item | Value |
|---|---|
| Build | $0 /mo |
| Ship | $50 /mo |
| Scale | $500 /mo |
| Enterprise | Custom |
| Build agent minutes included | 1,000 min /mo |
| Ship agent minutes included | 5,000 min /mo |
| Scale agent minutes included | 50,000 min /mo |
| Ship and Scale session overage | $0.01 /min |
| US local inbound overage | $0.01 /min |
| Third-party SIP, Ship | $0.004 /min |
| Third-party SIP, Scale | $0.003 /min |
| Example inference stack | $0.0479 /min |
Sources: livekit.io · Nov 1, 2026
What a LiveKit minute costs
Ship includes 5,000 min/month, then $0.01/min. Scale includes 50,000 min/month, then the same $0.01/min. Build includes 1,000 min/month and the cell does not state an overage rate. Concurrency is Build 5; Ship 20; Scale up to 600, starts at 50.
Per-second billing
LiveKit's blog, published Aug 24, 2026, says agent session minutes, recording, WebRTC connections, and SIP connections are charged per second, with a ten-second minimum. The change applies to all plans, including enterprise.
Telephony
US local inbound is $0.01/min after the included minutes on Ship and Scale. A US local number is $1/month after one free number. Third-party SIP is $0.004/min on Ship and $0.003/min on Scale, after each plan's included minutes.
Inference example
The pricing page's example stack totals $0.0479/min. The lines in that example are LLM $0.0014, STT $0.0075, TTS $0.0090, and observability. Model rows on the same page differ. Gemini 3.8 Live is listed at $0.0182/min there, which is LiveKit's price, not Google's token card.
Simulations and private links
Text simulations are $0.02 per turn and audio simulations are $0.05 per turn. The same rows still say free through Oct 31. This page was checked Nov 1, 2026. A private link on Scale is $50, plus $0.10 per GB.
Self-hosted
The Agents framework and the media server are open source. The Cloud page does not put a dollar price on running them yourself. You would pay for your own servers, models, and phone carrier. This page does not invent that bill.
Related pricing
Plivo Voice AI Agents is $0.03 per minute excluding telephony. The Plivo telephony section says SIP trunking works with LiveKit.
FAQ
Does LiveKit have a pricing calculator?
Yes. The pricing page includes an example stack that totals $0.0479/min. That total is the page's example (LLM, STT, TTS, and observability), not a flat rate for every model. Source: https://livekit.io/pricing (checked Nov 1, 2026).
What is the LiveKit cost per minute?
Agent session minutes are $0.01/min after the included minutes on Ship and Scale. Ship includes 5,000 min/month. Build lists 1,000 included minutes and does not print an overage rate. Source: https://livekit.io/pricing (checked Nov 1, 2026).
What is LiveKit self-hosted versus Cloud cost?
LiveKit Agents and the media server are open source. Self-hosting has no LiveKit Cloud session fee on the pricing page. You pay for your own servers, models, and telephony. This page does not invent a server bill.
How does LiveKit compare with Pipecat?
Pipecat is an open-source framework. LiveKit Cloud is a hosted plan: Build $0/month, Ship from $50/month, Scale from $500/month, plus $0.01/min after included agent minutes on Ship and Scale. Pipecat's own hosting cost is not a LiveKit price.
How does LiveKit compare with Vapi?
LiveKit Cloud agent session overage is $0.01/min on Ship and Scale. Vapi hosting is $0.05/min, with STT, LLM, and voice passed through. The bills are built differently.
How does LiveKit compare with Retell?
LiveKit Cloud agent session overage is $0.01/min on Ship and Scale. Retell AI voice infra is $0.055/min, plus LLM, voice, and phone. Those are not the same stack.
What is LiveKit Inference pricing?
Inference is billed from the model list after plan credits. The page's example stack is $0.0479/min. On that list, Gemini 3.8 Live is $0.0182/min. Those are LiveKit prices, not the model vendor's token card. Source: https://livekit.io/pricing (checked Nov 1, 2026).