Hardware
Hardware requirements are the same across deployment modes. The gateway is a routing layer, not a compute layer. Gateways are lightweight routing nodes. Minimum specs are sufficient for development and low-traffic deployments.Node type differences: AI or Dual gateways require Linux. Video gateways also support Windows via WSL2/Docker and macOS (build from source). On-chain gateways additionally require a public IP, static IP or domain, and ports 1935 (RTMP), 8935 (HTTP), and optionally 7935 (CLI).
Operating Systems
Not all operating systems support all node types.Networking Needs
Your gateway is a routing layer. Network quality directly determines job latency and stream reliability.- Off-chain
- On-chain
Off-chain gateways initiate outbound connections to orchestrators. They do not need to be publicly reachable.
Ports: Outbound to orchestrators on port 443 (or as configured). No inbound ports required.
On-Chain Requirements
On-chain deployment mode requires Ethereum infrastructure. Off-chain operators can skip this section.Gateways pay in ETH, not LPT. LPT is used for staking and governance by orchestrators - gateways never need LPT.
Off-Chain Requirements
Off-chain deployment mode removes all Ethereum requirements from your gateway. You need three things instead.Remote signer: The remote signer handles Ethereum payment signing on behalf of your gateway, so you do not need a funded wallet on the gateway node itself. Options: community signer (
https://signer.eliteencoder.net/ - free) or self-hosted for full control.Integration Notes
- Off-chain gateways supply orchestrator addresses directly via
-orchAddr. On-chain gateways discover orchestrators automatically through the Livepeer protocol. - AI or Dual node types require orchestrators running
ai-runnercontainers. Video node types work with any orchestrator advertising transcoding capabilities.
Technical Skills
Technical skills required depend on your deployment mode and stage.Minimum: comfortable with a terminal and Docker
Minimum: comfortable with a terminal and Docker
You can get an AI gateway running with:
- Basic Linux command line (navigate directories, run commands, read logs)
- Docker: pull an image, run a container, read container logs
- Ability to edit a configuration file or Docker Compose YAML
- An understanding of what a REST API is (for connecting your application)
On-chain: also need Ethereum basics
On-chain: also need Ethereum basics
Running a gateway in on-chain deployment mode requires:
- Creating and managing an Ethereum wallet
- Understanding the Arbitrum L2 network (not the same as Ethereum mainnet)
- Bridging ETH from Ethereum to Arbitrum
- Running
livepeer_clito allocate deposit and reserve funds - Monitoring your ETH balance and topping up when needed
Production: systems administration
Production: systems administration
For a production gateway handling production traffic, you additionally want:
- Linux systems administration (systemd services, log rotation, process monitoring)
- Basic networking (firewall rules, DNS, reverse proxy configuration)
- Docker Compose or equivalent for managing multi-container deployments
- Familiarity with Prometheus/Grafana or equivalent monitoring stack
Related Pages
Install with Docker
Quickest path to a running gateway. Docker handles dependencies automatically.
Install on Linux
Install the go-livepeer binary directly on your Linux host.
AI Gateway Quickstart
Step-by-step from zero to a running AI gateway, including Docker command and test request.
Community Install Methods
Ansible, Docker Compose, and community-maintained install scripts.