> ## Documentation Index
> Fetch the complete documentation index at: https://docs.daydream.live/llms.txt
> Use this file to discover all available pages before exploring further.

# Remote Inference

> Run Scope pipelines on cloud-hosted GPUs without local hardware

# Remote Inference (Beta)

<Warning>
  Remote Inference is currently in **beta**. The feature is free during the beta period, but functionality, availability, and pricing are subject to change.
</Warning>

The Remote Inference feature enables you to run Scope pipelines on cloud-hosted H100 GPUs provided by Daydream, eliminating the need for local GPU hardware. This means **Mac users** and anyone without a dedicated NVIDIA card can use Scope.

<Info>
  On Windows, only the **x64** version is supported - ARM-based Windows (e.g. Snapdragon laptops) is not supported at this time.
</Info>

***

## Setup

<Tip>
  You must sign in from **within the Scope desktop app** - not from the Daydream website. The sign-in option is located in the app's Account settings as described below.
</Tip>

<Steps>
  <Step title="Open Account settings">
    Launch the Scope app and navigate to **Settings** (gear icon, top-right) → **Account**.
  </Step>

  <Step title="Sign in with Daydream">
    In the Account panel, select **Sign in with Daydream**. A browser window will open for authentication — once complete, you'll be returned to the app.
  </Step>

  <Step title="Enable Remote Inference">
    After signing in, enable the **Remote Inference** toggle.
  </Step>

  <Step title="Start generating">
    Select your desired pipeline and press **Play**.
  </Step>
</Steps>

Once setup is complete, your Account settings should look like this:

<Frame>
  <img src="https://mintcdn.com/dd/NAcHocRhdDxotNsv/images/scope/guides/remote-inference/remote-inference-running.png?fit=max&auto=format&n=NAcHocRhdDxotNsv&q=85&s=ae7c370704b5e32925ae275da6521fd3" alt="Account settings with Remote Inference enabled and connected" width="1240" height="788" data-path="images/scope/guides/remote-inference/remote-inference-running.png" />
</Frame>

***

## Supported Features

Remote Inference has **full feature parity** with local Scope — everything you can do locally works in the cloud:

* **All pipelines** — Every pipeline available locally works remotely on H100 GPUs, including [LTX 2.3](/scope/guides/ltx-2-3) which is pre-bundled in the cloud image (no plugin install needed)
* **[LoRAs](/scope/guides/loras)** — Load LoRAs from CivitAI or Hugging Face; adjust strength in real-time
* **[Nodes](/scope/guides/plugins)** — Install and use community nodes (remote-approved nodes only)
* **[Workflows](/scope/guides/workflows)** — Import, export, and share workflows with full dependency resolution
* **Control interfaces** — OSC, MIDI, DMX, and other control protocols work as expected
* **Output interfaces** — Spout, Syphon, NDI, and other output protocols are fully supported

<Tip>
  When loading a workflow with LoRAs or nodes, the remote server automatically downloads and installs all dependencies. This may add 1-2 minutes to startup time.
</Tip>

***

## Usage Details

* **Session duration** - Each session provides up to 1 hour of GPU access. Upon disconnection, you may reconnect immediately to begin a new session. There is no charge and no limit on the number of sessions during the beta period.
* **Startup time** - Please allow approximately 2-3 minutes for the remote GPU to provision, followed by an additional 1-2 minutes for the selected pipeline to initialize and begin producing output.

***

## Known Limitations

<Note>
  These limitations are specific to the beta and may be resolved in future releases.
</Note>

* **[Krea Realtime](/scope/reference/pipelines/krea-realtime) pipeline switching requires an application restart** - to switch to `krea-realtime-video` from another pipeline, close the application, relaunch it, enable the Remote Inference toggle in the settings panel, select the Krea pipeline, and press Play.

***

## Troubleshooting

If you run into issues while using Remote Inference, Scope's built-in log panel and bug reporter can help you capture what went wrong and share it with us.

<Card title="Reporting Issues" icon="bug" href="/scope/guides/reporting-issues">
  Learn how to use the log panel, copy logs, and submit a bug report on GitHub
</Card>