Skip to main content
Page is under construction.

Check the github issues for ways to contribute! Or provide your feedback in this quick form
Orchestrators on Livepeer earn from two sources: ETH service fees paid by gateways for transcoding and AI inference work, and LPT inflationary rewards distributed each round to active staked operators.

Start here in 5 minutes

  • Prereqs: Active orchestrator setup, bonded LPT, and round/earnings visibility in Explorer
  • Time: 5 minutes
  • Outcome: Clear understanding of ETH vs LPT revenue and commission settings to tune
  • First action: Review Commission Parameters, then set target rewardCut/feeShare and verify your Reward() automation cadence

The Two Revenue Streams

1. ETH Service Fees (Demand-Driven)

Every time a gateway routes a transcoding or AI inference job to your orchestrator, it pays you in ETH. Payments flow via Livepeer’s probabilistic micropayment system — payment tickets that represent ETH obligations, redeemed on-chain periodically.
  • Transcoding fees — Earned for encoding video segments for live streams
  • AI inference fees — Earned for running AI pipeline jobs (text-to-image, LLM, image-to-video, etc.)
ETH fees are market-driven: they depend on how much work you win, which depends on your stake weight, pricing, performance, and capabilities.

2. LPT Inflationary Rewards (Protocol-Level)

Each round, the Livepeer protocol mints new LPT and distributes it to active orchestrators proportional to their stake weight. To claim your round’s reward, you (or an automated bot) must call the Reward() function on-chain.
  • Inflation rate is dynamic — it increases when less than ~50% of LPT is bonded, decreases when more than ~50% is bonded
  • You only earn inflation for rounds in which you call Reward() — missing a round means missing that round’s issuance
  • As of 02-March-2026, 10% of each round’s issuance goes to the Livepeer Treasury

Currency Summary

CurrencySourceHow You Earn It
ETHGateways paying for jobsWin transcoding + AI jobs; redeem winning payment tickets on-chain
LPTProtocol inflation (minted each round)Bond LPT, activate as orchestrator, call Reward() each round

Commission Parameters

When you register as an orchestrator, you set two commission parameters:

Reward Cut

The percentage of LPT inflationary rewards you keep as the operator. Your delegators receive the remainder, proportional to their stake.
Orchestrator LPT = Round inflation share × rewardCut
Delegator LPT    = Round inflation share × (1 - rewardCut) × (delegator stake / total stake)
Example: If your orchestrator earns 100 LPT this round and your rewardCut is 20%:
  • You keep 20 LPT
  • Delegators split 80 LPT proportionally to their stake

Fee Share

The percentage of ETH fees from jobs that is shared with delegators. The remainder you keep.
Delegator ETH = Job fee revenue × feeShare × (delegator stake / total stake)
Orchestrator ETH = Job fee revenue × (1 - feeShare)
Setting attractive commission rates helps attract more delegators, which increases your stake weight, which increases your inflation reward allocation and makes you more competitive for jobs. It’s a compounding competitive dynamic.

What Determines How Much Work You Win

Job routing in Livepeer is competitive. Gateways select orchestrators based on:
  • Stake weight — Higher total stake (your self-bond + delegators) means you’re higher in the active set
  • Capability matching — Do you support the requested pipeline, model, or resolution?
  • Performance history — Transcoding success rate, latency, uptime
  • Pricing — Are your fees competitive vs. other capable orchestrators?
  • Model warmth — For AI jobs, having models already loaded in GPU memory wins jobs over orchestrators with cold models

AI Inference vs. Transcoding Fees

As Livepeer expands into AI inference, the fee landscape is changing:
Video TranscodingAI Inference
UnitPer segment (few seconds of video)Per inference job
Fee levelLow (competitive, commoditised)Higher (GPU-intensive, model-dependent)
GPU requirementGPU recommended, not requiredGPU required
Job frequencyContinuous for active streamsPer-request
Model setupN/AWarm model = faster jobs = more wins
AI inference jobs command higher fees because they are more GPU-intensive. Orchestrators running NVIDIA GPUs with popular models loaded warm are positioned to capture a significant share of AI fee revenue as demand grows.

Payment Mechanics

Livepeer uses probabilistic micropayments to pay orchestrators efficiently at high frequency without paying on-chain gas for every segment:
  1. For each job, the gateway sends a payment ticket
  2. Each ticket has a probability of being a “winning ticket” worth a larger ETH amount
  3. When you receive a winning ticket, you submit it on-chain to redeem the ETH
  4. The expected value of tickets over many jobs equals the agreed service fee
This approach batches payments implicitly — instead of 1000 tiny on-chain transactions, you redeem periodic winning tickets. The Redeemer is a service that handles this on-chain redemption process.

Monitoring Your Earnings

The Livepeer Explorer shows:
  • Your current stake, delegators, and stake weight
  • Round-by-round reward history
  • Fee revenue over time
  • Active set status
For detailed metrics, Prometheus metrics are available from the go-livepeer node:

Prometheus Metrics Reference

Full list of metrics exposed by go-livepeer for monitoring.

Reward Calling Automation

Missing a round means missing that round’s inflation. Most production orchestrators automate Reward() calls:
  • go-livepeer flag: -reward=true enables automatic reward calling from your node
  • Ensure your node has sufficient ETH on Arbitrum to cover gas for the reward transaction
  • Monitor for missed rounds — the Explorer shows your reward calling history

Last modified on March 2, 2026