Overview
Livepeer is a full-stack decentralised infrastructure platform and marketplace for video streaming & AI. The video streaming software is underpinned by a network of actors that perform the work needed to compute, transcode & orchestrate video & AI jobs in the Livepeer network. Open-participation, distributed infrastructure systems like Livepeer aren’t just software systems - they’re multi-layered digital ecosystems that require careful design of economic incentives, governance structures, and technical architecture. From a marketplace perspective, Livepeer has:- A demand-side agent (an app, platform, or operator) which brings video or AI jobs;
- A supply-side agent (Orchestrators and their GPU capacity, optionally backed by separate worker processes) which performs compute and other work;
- A protocol which provides the governance, payment primitives and reward system to coordinate open participation and make the market viable at scale.
- A hardware layer of GPU nodes (orchestrators) that perform the compute work
- A network layer of
go-livepeernodes that handle job routing, verification and payment; - A software layer of services built on top of the network layer (e.g. Studio, Daydream, ComfyStream) that provide additional functionality and services to the network.
- A protocol layer of smart contracts that enforce the rules of the network.
- An economics layer of tokenomics and game theory that aligns incentives.
Livepeer Protocol
The Livepeer Protocol is the underlying on-chain logic that enforces the mechanisms, incentives and rules to ensure the security, reliability, cooperation and coordination of these decentralised actors for the desired network outcomes (video streaming & AI pipelines). The protocol additionally defines the economic mechanisms and governance rules that enforce staking, bonding, slashing, and reward rules on-chain, while off-chain nodes and services handle the actual compute jobs and job routing. Protocol Services:- staking
- delegation
- inflation & rewards
- orchestrator selection
- slashing
- probabilistic payments
- verification rules
Livepeer Network
The network is the actual running system and execution layer - the mesh of hardware and software nodes performing work:- Orchestrators (GPU nodes)
- Transcoders / Workers
- Gateways (aka Broadcasters)
- Verification processes
- Job routing
- Real-time AI & video compute
- Job routing
- Verification
- Payment
- AI inference
- Video Transcoding
- Real-time AI Video Streaming
Protocol vs Network
The protocol governs. The network performs the work.| Layer | Description |
|---|---|
| Livepeer Protocol | On-chain crypto-economic incentives & coordination; staking; payments. |
| Livepeer Network | Off-chain nodes performing real-time work (transcoding, inference, routing). |
| Relationship | The network runs the compute; the protocol governs, secures, and pays it. |
On-chain vs Off-chain
- Livepeer protocol = Arbitrum One (L2)
- Livepeer token (LPT) = Ethereum mainnet (L1)
- Livepeer network (GPU nodes + gateways) = off-chain execution layer
Livepeer Nodes, Actors & Roles
A Livepeer actor is any role or entity that participates in the Livepeer protocol or network and performs actions defined by the system. In Livepeer architecture, “actor” is a formal category used to describe participants with distinct responsibilities, incentives, and interactions. Actors are fundamental to describing how the network functions end-to-end.Orchestrators
GPU nodes that perform transcoding / AI jobs, earn fees + staking rewards.Roles: Perform Compute Operations, participate in governance
Gateways
Clients submitting jobs (video or AI) into the network. The users and builders of the network.Roles: Submit jobs, route work, manage payment flows, connect to orchestrators
Delegators
LPT holders who stake to an orchestrator.Roles: Stake LPT, earn rewards, participate in governance and voting
Platforms? Workers?
Services built on top of the Livepeer network (not directly part of the network).Roles: Build on top of the network, provide developer-friendly and end-user services
- Transcoders: Performs video transcoding operations.
- AI Workers: Executes AI inference tasks (text-to-image, LLM, video-to-video, etc.)