Hugging Face Spaces just got a lot more fun. The platform known for hosting machine learning demos now supports playable Unity games, and the team published a step-by-step guide showing exactly how to pull it off. The tutorial walks through building a simple game called Huggy — an AI dog that catches sticks you throw — and deploying it as a WebGL build inside a Static HTML Space.
The process is more straightforward than you might expect. You create a Space using the Static HTML template, clone it locally, switch your Unity project’s build target to WebGL, and disable compression in the Publishing Settings. Hugging Face even provides a custom Unity WebGL template that adjusts the game’s appearance to fit the Space layout. From there, it’s standard git workflow: track the build files with Git LFS, commit, push, and refresh.
What’s notable here is the signal it sends about platform direction. Spaces was built for ML demos — Gradio apps, Streamlit dashboards, model inference widgets. Hosting Unity games stretches that definition considerably. It suggests Hugging Face is thinking beyond the traditional ML researcher audience and courting game developers, creative coders, and anyone building interactive browser experiences. The Discord invitation at the end of the tutorial reinforces that: they’re actively building a community around games on the platform.
There’s a practical angle too. WebGL builds are notoriously finicky about hosting — compression settings, MIME types, and large binary files trip up even experienced developers. Hugging Face handling the infrastructure removes real friction. The Git LFS step is crucial; Unity build folders routinely exceed GitHub’s 100MB file limit, and skipping LFS means failed pushes. The tutorial covers it, but plenty of first-timers will still hit snags there. Still, the barrier to shipping a browser-playable game just got meaningfully lower, and that’s worth paying attention to.
