跳转到主要内容
Page is under construction.

Check the github issues for ways to contribute! Or provide your feedback in this quick form
Livepeer Orchestrators are both general purpose GPU compute providers, as well as economic actors that stake LPT, earn rewards, and vote in governance. They also host and provide AI models & inference endpoints (including BYOC) which power novel services like AI avatars and agent workloads. Additionally, they may also manage delegated LPT and can have significant input into the direction of the network by voting as a block in governance with their delegators.
Orchestrator Dual Role as defined by Andrew Macpherson in his

​
General Compute Functions

Orchestrators perform two core compute functions on the Livepeer network: transcoding and AI inference. They do this by running a node that is connected to the network and performing the work that is assigned to it.

​
Transcoding

转码是将视频从一种格式转换为另一种格式的过程。 在 Livepeer 网络上的协调者通过运行连接到网络并执行分配给它的任务的节点来执行转码。

​
AI 推理

AI 推理是运行模型以进行预测的过程。在 Livepeer 网络上的协调者通过运行连接到网络并执行分配给它的任务的节点来执行 AI 推理。

​
自定义工作负载

协调者还可以在 Livepeer 网络上运行自定义工作负载。

​
AI 管道功能

​
BYOC

​
AI 模型

​
委托与治理服务

协调者还可以通过将 LPT 代币委托给其他协调者来获得奖励。

​
由协调者提供的示例服务

Livepeer 向实时 AI 视频的转型意味着协调者除了基本的转码外，还需要运行 AI 代理、虚拟化身流和互动体验。
  • 交互式 AI 视频
    • 白日梦
  • 实时化身与虚拟代理
    • 化身
    • 莎拉
  • 工作流自动化与代理工作流
  • 交互式直播体验（AR/VR、游戏）
    • 元宇宙
    • 直播游戏
    • 互动直播活动
Last modified on March 1, 2026