Sanctum

A private AI assistant that enables users to run full-featured open-source LLMs on their devices

Sanctum is a private AI assistant that enables users to run full-featured open-source Large Language Models (LLMs) on their devices, ensuring data privacy and security. By processing information locally, Sanctum eliminates the need for cloud-based systems — thus reducing the risk of data breaches and unauthorized access. This approach can be particularly beneficial for professionals handling sensitive information, such as legal documents, financial data, and/or personal health records.

Sanctum integrates with Hugging Face, allowing users to access and interact with thousands of GGUF models directly through the platform. This integration means seamless downloading and utilization of various models on both PC and Mac, enhancing the tool’s applicability across different tasks and industries.

Sanctum also allows import and interact with various document formats, including PDFs, DOCX, TXT, and code files. Users can chat with AI to analyze or summarize these documents without compromising privacy, as all processing occurs locally on the device.

Sanctum provides flexible pricing plans to accommodate different user needs. The platform offers a free version with limited features, while paid plans unlock additional functionalities and support. This flexibility is there to ensure that both individual users and organizations can find a plan that aligns with their requirements and budget. Nice!

Homepage Screenshot 📸

What are the key features? ✨

Easy Setup: Download and run the app instantly on Mac or Windows with minimal configuration needed.
Privacy-First Design: All data processing and storage use AES-256 encryption on-device so nothing leaves your machine.
HuggingFace Integration: Browse, check compatibility, and download thousands of GGUF open-source models directly in the app.
Local PDF Chat: Upload PDFs or DOCX files to ask questions, get summaries, and interact privately with documents.
Offline Operation: Work fully offline after initial model download with no internet required for AI tasks.

Who is it for? 🤔

Sanctum suits privacy-conscious individuals, professionals dealing with sensitive data (health, finance, legal, or business documents), developers testing open-source models, and anyone wanting reliable AI assistance without cloud dependency or data exposure risks. Its especially valuable for offline scenarios or secure environments where control over information matters most.

Examples of what you can use it for 💡

Privacy-focused professional: Analyze confidential reports, call notes, or client data securely without external servers.
Personal user: Summarize personal documents, track wellness logs, or brainstorm ideas while keeping everything local.
Developer or coder: Experiment with different open-source LLMs, get code suggestions, or review snippets privately.
Researcher or student: Query and summarize academic PDFs or long readings offline with full data control.
Business user: Review internal financials, user feedback, or sales materials without risking data leaks.

Pros & Cons ⚖️

Top-tier privacy
Easy local setup
Offline capability
Model variety

Hardware-limited speed
Current OS restrictions

FAQs 💬

What platforms does Sanctum support?

It currently runs on Mac and Windows 10+, with Linux support planned soon and mobile versions in development.

Is Sanctum really private?

Yes, all processing happens locally with AES-256 encryption; your data, prompts, and files never leave your device.

Do I need internet after setup?

No, once models download you can use chat, document features, and everything else fully offline.

How do I choose a model?

The built-in AI matching engine shows compatible GGUF models from Hugging Face and suggests options based on your hardware.

Can I chat with my PDFs?

Yes, upload PDFs or DOCX files to ask questions, request summaries, or discuss content privately.

Is Sanctum free to use?

The app is free to download and use with open-source models; no subscription is required.

What hardware do I need?

It works on standard laptops, but a good GPU speeds up larger models significantly; smaller quantized models run fine on basic hardware.

How does it compare to Ollama?

Sanctum offers a more user-friendly GUI with integrated model browsing and stronger privacy defaults, while Ollama provides more customization for advanced users.