BC ModelVault — Hugging Face Downloader and AI Model Manager

Stable version: 1.0.1 · Windows · Python 3.10+

Official repository: BalzanoConsulting/BC-ModelVault

BC ModelVault is a Hugging Face downloader for Windows and a local AI model manager. Download complete Hugging Face models, datasets, and Spaces from a URL or repository ID, including private and gated repositories. Downloads are resumable, pinned to an exact revision, checked with trusted SHA-256 hashes when available, and indexed in a searchable local AI asset library.

Use BC ModelVault as a graphical alternative to manual huggingface-cli downloads when you need download history, restart recovery, integrity validation, duplicate detection, and organization for LLM, checkpoint, LoRA, VAE, and ControlNet files.

BC ModelVault vs Hugging Face CLI

The Hugging Face CLI is excellent for scripts, terminals, CI pipelines, and users who want direct command-line control. BC ModelVault builds on the official huggingface_hub library and adds a desktop workflow for model collections:

Capability	BC ModelVault	Hugging Face CLI workflow
Graphical Windows interface	Built in	Terminal commands
Restart recovery	Persistent SQLite queue restores interrupted jobs	Resume must be managed from commands and cache state
Download overview	Progress, speed, ETA, pause, history, and status in one interface	Terminal output per command
Integrity status	Explicit verified, unverified, or mismatch state	Requires interpreting download/cache results or additional checks
AI model catalog	Searchable local library for LLM, checkpoint, LoRA, VAE, and ControlNet	Not a catalog interface
Model classification	Automatic type detection with visible confidence and manual override	Requires user scripts or external tooling
Duplicate detection	Cryptographic hash comparison across downloaded assets	Requires separate commands or scripts
Metadata browsing	Author, license, revision, date, format, size, and tags together	Available through separate commands/API calls
Private/gated repositories	Multiple saved token profiles in the interface	Token/login commands
Diagnostics	Rotating redacted logs and token-free report export	Depends on shell logging and user configuration

BC ModelVault is not intended to replace the CLI for automation. It is intended to make repeated Hugging Face model downloads, verification, recovery, and local asset management easier for desktop users.

Hugging Face downloader features

• Download public, private, and gated Hugging Face repositories on Windows.
• Resume interrupted model downloads across application restarts.
• Cancel transfers cooperatively while preserving reusable partial cache data.
• Retry transient network/server failures with bounded exponential backoff.
• Run 1, 2, 4, or 8 concurrent Hugging Face file downloads.
• Apply an optional aggregate download speed limit in MB/s from the Settings page.
• Pin every download to the verified Hugging Face commit revision.
• Validate Hugging Face LFS files using published SHA-256 digests.
• Manage LLM, checkpoint, LoRA, VAE, and ControlNet assets in a local catalog.
• Search and filter downloaded AI models by type and integrity status.
• Compare installed and current Hugging Face revisions without implicit overwrites.
• Detect duplicate AI model files by cryptographic hash.
• Install downloaded assets into multiple validated ComfyUI profiles using reviewed routing plans.
• Choose copy, hard-link, or symbolic-link installation with safe copy fallback.
• Keep tokens out of rotating logs and exported diagnostic reports.

Download BC ModelVault for Windows

Download the latest BC-ModelVault.exe

The portable executable does not require Python. It is published as a GitHub Release asset together with SHA256SUMS.txt. Build intermediates under build/ and dist/ are not part of the source repository. Windows SmartScreen may display a warning when a release is unsigned; verify the checksum before selecting More info → Run anyway.

The repository is intended for source code and development. End users should download the executable or the Windows ZIP from Releases rather than browse the Python tree.

Run from source

Requirements: Windows and Python 3.10 or later.

python -m pip install -r requirements.txt
python -m bc_modelvault.app

How to download a Hugging Face model

1. Paste a Hugging Face URL or an owner/repository identifier.
2. Choose the destination folder.
3. For private or gated repositories, add a Read token from Account → Add token….
4. Select Check package to validate files, space, and Windows filenames.
5. Select Download package. The transfer can be paused and resumed.

Content is stored under owner--repository and pinned to the verified commit. After downloading, BC ModelVault validates file sizes and trusted Hugging Face LFS SHA-256 digests, then adds the repository to the local AI model library. Tokens are managed by the official huggingface_hub library and are never stored in the project or written to logs.

Documentation

• User guide
• Architecture and development
• Implementation roadmap
• Windows build
• Installation
• Release policy
• Changelog
• Contributing
• Security
• Support
• Known issues
• End-of-life policy
• Compliance and provenance
• Third-party notices

Project structure

• bc_modelvault/app.py: user interface and event coordination.
• tests/: automated unit and integration tests.
• bc_modelvault/core.py: URLs, metadata, and file validation.
• bc_modelvault/progress.py: Hugging Face download progress.
• bc_modelvault/localization.py: languages and preferences.
• bc_modelvault/theme.py: light and dark themes.
• bc_modelvault/tray.py: Windows notification area integration.
• languages/: Italian and English catalogs.
• img/logo.png: transparent Balzano Consulting logo used by the window, tray, and executable.
• img/icons/: optional PNG icons for sidebar and actions; missing icons use text fallback.

Optional interface icons

Place transparent PNG files in img/icons/ to enable icons without changing code. The supported filenames and sizing recommendations are documented in img/icons/README.md. If an icon is absent, BC ModelVault keeps the full text label, so accessibility and navigation are not affected.

Current status

BC ModelVault currently provides a persistent Hugging Face download queue, restart recovery, concurrent workers, SHA-256 validation, model metadata, automatic AI asset classification with confidence levels, manual classification overrides, a searchable local model library, duplicate detection, secure diagnostics, and reviewed ComfyUI installation with multiple profiles, conflict handling, and optional links.

License

Copyright © 2026 Balzano Consulting. All rights reserved. This project is distributed under the PolyForm Noncommercial License 1.0.0. Use, modification, and redistribution are allowed only for noncommercial purposes and require preservation of the license and attribution. Commercial use requires separate permission from the copyright holder. See LICENSE, NOTICE, and third-party notices.