LiveKit
The open-source WebRTC server behind modern voice AI - with the 10,000-port UDP problem collapsed to one.
One-click deploy, from $7/mo on a Miget plan.
LiveKit became the substrate of the voice AI wave: a single Apache-2.0 Go binary doing WebRTC rooms, tracks, and selective forwarding, with the Agents framework ecosystem on top. Self-hosting it historically meant a 10,000-port UDP range - hostile to any platform.
This template uses the config that fixes that: ALL media muxes over one UDP port (7882), with one TCP fallback (7881) and the websocket signal on :5000 behind the ordinary HTTPS ingress. Two custom public ports, same numbers inside and out, and WebRTC works.
Pair it with the livekit-agents template and you have a complete self-hosted voice-agent platform: this server moves the audio; the agent worker brings ElevenLabs, Deepgram, and your LLM gateway.
Upstream project: LiveKit
#what you get
- WebRTC rooms, tracks, SFU - the full LiveKit server
- Single muxed UDP port instead of a 10,000-port range
- Websocket signal behind normal HTTPS (:5000)
- Config via one env var (LIVEKIT_CONFIG) - no file mounts
- Apache-2.0; the same server LiveKit Cloud runs
- Agents ecosystem: voice AI workers join rooms natively
#topology
| Service | Role | Public |
|---|---|---|
| livekit | signal :5000 (HTTPS ingress) + ICE 7881/tcp + media 7882/udp | yes + two custom ports |
#miget sizing
// this stack needs
1 GiB RAM · 0 GB disk · 1 service
Scaling is CPU and bandwidth, not RAM - one node carries small-production voice workloads comfortably. Redis enters only for multi-node.
Hobby - recommended fit
$7/mo
1 vCPU · 1 GiB · 25 GiB disk
Headroom for your own apps: 2 GiB at $13/mo
Professional - production
$22/mo
1 vCPU · 2 GiB · 10 GiB disk
Dedicated resources, production SLOs - plan details
One Miget plan is a fixed pool of compute - the whole stack (managed databases included) deploys inside it, and anything left over runs your other apps. No per-service or per-seat math.
#vs. other PaaS
Estimated monthly cost of running this exact stack (1 GiB RAM, 0 GB disk, 1 container) elsewhere, from published June 2026 rates.
| Platform | Est. monthly | Notes |
|---|---|---|
| Miget ★ | $7 flat | compose stacks first-class: one deploy, dedicated vCPU, managed Postgres/Valkey, volumes and TLS all included in the plan |
| Heroku | ~$50 | no volumes; nothing between 1 GB ($50) and 2.5 GB ($250) dynos - 2 GB containers cost far more than shown |
| DO App Platform | ~$17 | no persistent volumes - stateful containers need managed DBs/Spaces (base $5 Spaces included here) |
| Render | ~$13 | per-service instances (0.5 GB $7, 2 GB $25) - every container is its own paid service |
| Railway | ~$10 | usage-based ($10/GB RAM-mo); vCPU billed separately at $20/vCPU-mo on top |
| Fly.io | ~$6 | cheapest sticker price - but burstable shared CPUs (1/16 core; dedicated vCPUs cost ~2-3×), no compose deploys (one app per container, manual wiring), managed DBs billed extra |
Estimates assume RAM fully allocated at published on-demand rates - and sticker price isn't the whole comparison: the cheaper rows buy burstable shared CPUs, per-service wiring instead of a compose deploy, and managed databases billed separately. Heroku and DO App Platform have no persistent volumes at all - stateful stacks like this one need workarounds there.
#deploy it
On Miget
- Create a Compose Stack in app.miget.com pointing at the templates repository
- Set the stack path to
livekit -
Set the required variable:
LIVEKIT_API_KEY / LIVEKIT_API_SECRET, the key pair your apps mint access tokens with
- Deploy. Miget layers
compose.miget.yaml(RAM, privacy, volumes, managed services) automatically
Locally first?
Every template is portable, vanilla Docker Compose - the Miget overrides are ignored locally:
git clone https://github.com/deployable-sh/stacks
cd miget-compose-templates/livekit
docker compose up -dSame files, same behavior. The template README covers connection strings and scaling notes.
#faq
How does WebRTC work on an HTTP-first platform?
Three ports, deliberately: the signal is a websocket on the normal HTTPS domain, and media rides ONE muxed UDP port (7882) with a TCP fallback (7881) - both exposed as custom public ports with matching numbers, since ICE advertises them. Clients on hostile networks fall back to TCP automatically.
Why self-host instead of LiveKit Cloud?
Cloud is excellent and usage-priced - the honest default for spiky workloads. Self-hosting wins for always-on voice agents (predictable cost), data locality (audio never transits a third party), and pairing with the in-project agent worker at LAN latency.
What do clients connect with?
Access tokens minted from your API key/secret by any LiveKit server SDK - one function call in your backend. The client SDKs (web, mobile, Unity) then connect to wss://your-domain and join rooms.
Ship LiveKit today
One compose stack, 1 GiB of RAM, from $7/month flat, and it runs on your laptop with the same files.