Skip to content

Get Started

Welcome to FlashDreams!

This page will guide you into running your first world-model.

We will use NVIDIA OmniDreams, an interactive driving world model, as the example:

OmniDreams interactive driving demo.

Install

First, clone FlashDreams: 

git clone https://github.com/NVIDIA/flashdreams.git
cd flashdreams

Now we need to install uv, FlashDreams's package manager. Installation instructions here.

With uv installed, clone the repository and synchronize uv to the OmniDreams workspace via: 

uv sync --package flashdreams-omnidreams --extra interactive-drive --inexact

Most runs need a Hugging Face token for model/example-asset downloads. For OmniDreams, use a token with read access to nvidia/omni-dreams-models and nvidia/omni-dreams-scenes:

export HF_TOKEN=<your-hf-token>

Run your first model

Launch the interactive driving demo in a browser over WebRTC: 

uv run --no-sync flashdreams-run-v2 \
    interactive-drive-omnidreams --mode webrtc \
    --host 0.0.0.0 --port 8089
# Open in browser: http://<server-ip>:8089/

The first launch spends several minutes loading checkpoints and compiling kernels.

See our How to Run a Model for more information on the CLI run command to tune the arguments to your liking.

Install one integration

Install only the model package you need with --package. Add --extra dev when working on its code:

uv sync --package flashdreams-omnidreams --extra dev --inexact

Use the distribution name from the integration's pyproject.toml. See Models for available integrations, requirements, and setup commands. See Testing for test commands.

Where to next