Build with Reactor

These docs cover the reactor CLI and deployment, and describe features that are rolling out to partners. Contact us for access. Writing the model itself is documented at docs.reactor.inc/deploy.

Reactor is infrastructure for real-time video and world models. You write a Python model, test it locally, and deploy to Reactor’s global GPU network. We handle WebRTC streaming, session management, and delivery.

How it works

Install the CLI

Install the reactor CLI and you are set. The runtime ships inside the image the CLI builds for you, so nothing else installs on your host.

curl -fsSL https://reactor.inc/install | bash

Scaffold a workspace

Use the reactor CLI to scaffold a Docker-based workspace. The generated reactor.yaml declares everything Reactor knows about the model: registration metadata in model:, runtime entrypoint in runtime:, and the image in build:.

reactor init my-model
cd my-model

reactor.yaml

model:
  name: my-model
  version: v0.0.1

runtime:
  import: model:MyModel
  config: config.yaml

build:
  runtime_version: "3.2.5"

Build your model

Subclass ReactorModel, load your weights in load(), and emit frames from run(). The runtime streams them to clients over WebRTC.

class MyModel(ReactorModel):
    async def run(self):
        while True:
            await self.connected.wait()
            while self.connected.is_set():
                frame = self.pipe.forward(prompt=self.prompt)
                await self.emit(MyOutput(main_video=frame))

Writing the model is documented at docs.reactor.inc/deploy.

Test locally

reactor run builds the workspace image and starts it on port 8080. Connect from the Reactor Sandbox in Local (Direct) mode to see frames streaming.

reactor run

Deploy

Run the deploy commands from inside the workspace. The CLI reads the model name and release tag from reactor.yaml, so you register once, then publish and deploy each release. Your model is live on production GPUs in under 3 minutes.

reactor auth login
reactor model register
reactor model publish
reactor model deploy

Connect clients

Use the JavaScript or Python SDK to stream your model’s output to any application.

<ReactorProvider modelName="my-model" autoConnect={true}>
  <ReactorView className="w-full aspect-video" />
</ReactorProvider>

Why Reactor

Sub-50ms streaming

Frames delivered over WebRTC as they are generated. Client inputs received live.

Stateful sessions

Your model holds state across a whole session, with hooks for every client that joins or leaves.

Global GPU network

Nodes in every major region. A client in Tokyo connects to a GPU in Tokyo.

No transport code

You never touch WebRTC, WebSockets, or video encoding. Reactor handles it.

Live in minutes

Publish and your model is running on production GPUs in under 3 minutes.

You own your model

Your weights, your inference logic. Reactor never accesses or trains on your data.

Get Started

CLI & Platform

Build a Model

Media & Transport

Releases

Build with Reactor

How it works

Why Reactor

Sub-50ms streaming

Stateful sessions

Global GPU network

No transport code

Live in minutes

You own your model

Get started

Quickstart

Build a Model

Install the CLI

Deployment

​How it works

​Why Reactor

Sub-50ms streaming

Stateful sessions

Global GPU network

No transport code

Live in minutes

You own your model

​Get started

Quickstart

Build a Model

Install the CLI

Deployment

How it works

Why Reactor

Get started