What is ModelHub?

ModelHub is a native macOS menu bar app designed for developers working with local large language models (LLMs). It acts as a centralized discovery and management layer for Hugging Face models, letting you browse, download, and organize models directly from your menu bar. Once downloaded, you can run those models with Ollama, MLX, LM Studio, llama.cpp, or any other tool you already use — without juggling browser tabs, terminal commands, or scattered local folders. It's a lightweight (~4 MB) companion that fills the gap between model discovery and local execution.

Who it's for

• Developers using local LLMs — who want to find and download Hugging Face models without leaving their workflow.
• macOS power users — who prefer a menu bar interface over terminal commands or browser-based model browsing.
• Multi-tool practitioners — who run models across Ollama, MLX, LM Studio, and llama.cpp and need a unified way to manage their model library.

Key features

One-click model discovery and download

Browse Hugging Face models directly from the menu bar, then download them with a single click. ModelHub replicates the official Hugging Face cache layout — blobs, snapshots, and refs — so every download is byte-identical to one fetched with huggingface-cli. No migration needed if you switch tools later.

Cache-compatible with zero lock-in

ModelHub reads and writes the standard ~/.cache/huggingface/hub/ directory structure. Models you download work immediately with Ollama, MLX, LM Studio, llama.cpp, and Hugging Face Transformers. You can uninstall ModelHub at any time — your models stay intact and ready for any pipeline.

Unified model library management

See all your local models in one place, regardless of which tool downloaded them. Remove models, check disk usage, and keep your library organized without hunting through random folders or terminal commands.

What stands out

ModelHub is the missing discovery and management layer around Ollama, MLX, and LM Studio.

While those tools handle model execution well, none of them provide a unified interface for finding, downloading, and organizing models from Hugging Face. ModelHub fills that gap by living in your menu bar — always accessible, never in your way. It respects the standard Hugging Face cache layout, so there's no proprietary format or vendor lock-in. You get the convenience of a GUI without sacrificing compatibility with the tools you already rely on.

Worth checking out if…

You work with local LLMs on macOS and find yourself switching between browser tabs, terminal commands, and multiple model folders. ModelHub is especially useful if you use more than one local LLM tool (Ollama, MLX, LM Studio, llama.cpp) and want a single place to discover, download, and manage your models. It's also a great fit if you prefer menu bar utilities over full-window apps and value zero-lock-in compatibility with the Hugging Face ecosystem.