Hugger provides a one-click solution for archiving HuggingFace models to a self-hosted server, designed to help users maintain access to models that may become unavailable due to regulatory changes, takedowns, or licensing updates. The tool addresses the risk of losing access to important models by allowing individuals or organizations to keep their own local copies, ensuring continued availability and reproducibility.
The platform consists of a browser extension available for Chrome, Firefox (including Firefox on Android), and Edge, which connects directly to a user's own hugger server. This server is a lightweight Python application featuring both a web interface and an API, capable of downloading full model snapshots to disk. Users can expose the server to the internet, and it supports login and token-based authentication. There is also mention of a NixOS module for simplified deployment. The extension is configured by pasting in a token, which is securely held by a background worker so that the token is not exposed to the web page itself.
From any HuggingFace model page, users can archive a model with a single click. Hugger downloads the snapshot and records the specific commit revision, supporting reproducibility for research and audits. The extension interface indicates when a model has been archived, provides an option to remove the archive, and offers an update feature when a newer revision of the model is released. All communication occurs between the extension and the user’s own server, with no data sent to third parties.
Hugger is positioned as a tool for those who rely on HuggingFace models and require control over their model assets, including researchers, developers, or organizations with air-gapped environments or specific data-residency needs.
In the Other infrastructure space, hugger takes a focused approach. It focuses on ensuring reliable, local access to HuggingFace models by archiving them before they become unavailable. hugger is an open-source project aimed at AI researchers and developers using HuggingFace models. The project is open source (Open Source). It runs on the web, embeddable surfaces, Linux, Windows, and macOS, and it can be self-hosted.
hugger first shipped in 2026. Across PulseGate's embedding index, hugger has few near neighbours, marking it as relatively distinct. Among its 5 catalogued features are model archiving, self-hosted server, and browser extension.
Latest indexed changes and source events
hugger — one-click model archiver discovered by the PulseGate indexer
Other apps tracked under the same category.
