Cloud GPU for Unreal Engine: render the sequence, stream the build, previs the film

Build the level on your own machine. When the shot is locked, open an RTX 4090 or RTX 5090 on CloudGPU and let Movie Render Queue run the final frames, billed by the minute, from one balance. SSH session. Your project, your Unreal Engine install, your output folder. Which cards you can start right now is shown on /deploy.

RTX 4090 / RTX 5090 tiers · per-minute billing · one balance for GPU + model APIs · card or USDT; see /payments

Open a GPU · Read the MRQ guide

Who it’s for

1. Movie Render Queue artists

You’ve blocked the cinematic, the camera is locked, and your laptop fan is the loudest thing in the room. Push the project to a rented GPU, run MRQ headless with -RenderOffscreen, and pull back EXR or PNG sequences while you keep working locally.

2. Pixel Streaming prototypers

You need a stakeholder to click around your packaged build in a browser: a configurator, a walkthrough, a pitch demo. Run the build on an NVENC-capable NVIDIA GPU and stream it. Sized for prototypes and small-audience demos, not large-scale public launches. Guide: Pixel Streaming on a cloud GPU.

3. AI previs with a Blender or UE greybox

Block the scene as a greybox in Blender or Unreal, render rough camera passes, then feed them into open video models like Wan or MiniMax H3 to explore look and mood before you commit to a full build. Same balance, same evening. See the full film pipeline: Cloud GPU for AI video production.

What you get

RTX cards for real-time and offline rendering

Choose an RTX 4090 or RTX 5090 tier for the job. Lumen and Nanite previews, high sample-count anti-aliasing passes in MRQ, and NVENC hardware encoding for Pixel Streaming all run on the same class of card.

Pay for the render, not the idle week

Billing is per minute, so a two-hour overnight sequence is a two-hour session. Re-render a shot after a note, destroy the instance, done. Billing rules are on /pricing.

Headless over SSH

SSH in and launch the editor binary from the command line with your queue. Copy your files out before you destroy the instance.

One balance for GPU time and model APIs

The same prepaid balance covers GPU instances and OpenAI-compatible model APIs, which is handy when your previs loop mixes Unreal renders with AI video or image generation.

Your files stay your workflow

Bring your project via git, rsync, a Perforce client, or a zip upload. Render to the instance and pull results back the way you already move files.

How it works

  1. Prepare locally. Save your Level Sequence and an MRQ queue or config asset. For Pixel Streaming, package a build with the Pixel Streaming plugin enabled.
  2. Open a GPU. Go to /deploy and pick an RTX 4090 or RTX 5090 tier.
  3. Bring your own Unreal Engine install/account. Install Unreal Engine on the instance using your own Epic account, or copy your own engine build. CloudGPU provides the GPU machine; it does not bundle Unreal Engine.
  4. Upload the project. Sync your project and assets to the instance.
  5. Render or stream. Run MRQ headless (-RenderOffscreen), or launch your packaged build with Pixel Streaming.
  6. Pull results and destroy. Download frames or logs, then destroy the instance so billing stops with it.

CloudGPU is GPU rental: a machine you control over SSH. It is not an Epic Games product, not an official Unreal Engine cloud-render plugin, and not affiliated with or endorsed by Epic Games.

FAQ

Do I need my own Unreal Engine install/account?

Yes. Bring your own Unreal Engine install/account: sign in with your Epic account on the instance, or deploy your own engine build. Your use of Unreal Engine stays between you and Epic’s terms. CloudGPU rents the GPU machine only.

Is CloudGPU an official Epic or Unreal Engine partner?

No. CloudGPU is an independent GPU rental service. We don’t offer an Epic-certified cloud or an official Unreal plugin: you get a GPU instance and run Unreal Engine on it yourself.

Can I run Movie Render Queue without a monitor?

Yes. Launch the editor in command-line mode with -RenderOffscreen and your queue or config asset. Step by step: MRQ on a rented CloudGPU.

Does Pixel Streaming work on CloudGPU cards?

Pixel Streaming needs an NVIDIA GPU with NVENC; RTX 4090 and RTX 5090 cards include it. We suggest it for prototypes, reviews and small demos. Guide: Pixel Streaming on a cloud GPU.

RTX 4090 or RTX 5090?

Pick the tier that fits your scene’s VRAM needs and deadline. Heavy Nanite scenes, large textures and high-resolution MRQ passes benefit from more headroom. See Rent an RTX 5090 and /pricing for current tiers.

How do I get my project onto the GPU?

Any way you’d move files to a remote Linux machine: git, rsync/scp, a Perforce client, or an archive upload. Keep the Derived Data Cache on the instance for repeat renders while it is running.

Can I combine Unreal renders with AI video?

Yes, that’s the previs workflow. Render greybox camera passes from UE or Blender, then run open models such as Wan or MiniMax H3 on the same balance. See Cloud GPU for AI video production.

How do I pay?

Card or USDT; see /payments.

Your shot is locked. Open the GPU.

Render the sequence tonight, stream the build tomorrow morning.

Open a GPU

Card or USDT; see /payments.

Open a GPURead the MRQ guide