Local LLM Chat
Run open-weight models — Llama, Qwen, Mistral, Gemma — at any size your hardware can handle. Quantized 7B to 70B+, fully offline.
100% offline · Zero telemetry · Your machine, your data
Your private AI workstation. Fully offline.
Vault AI runs powerful language models, document Q&A, OCR, transcription, and PII redaction entirely on your own machine. Built for legal, medical, financial, and security-conscious professionals who can’t send data to the cloud — ever.
Available for
Windows
macOS
No data leaves your device. Network access is opt-in for updates only.
Air-gap ready. Once activated, runs fully offline — no internet required.
30-day money back. Try it risk-free.
Features
Everything you need. Nothing phoning home.
A complete AI workstation that runs entirely on your hardware.
Document Q&A (RAG)
Index thousands of PDFs, Word docs, and notes. Ask questions, get cited answers — all retrieval and inference happens on-device.
PII Redaction
Detect and redact names, emails, IDs, addresses, and custom patterns before sharing. Works on text, PDFs, and images.
Offline OCR
Extract text from scanned documents, screenshots, and photos in 80+ languages. Searchable PDFs, no upload required.
Meeting Notes (ASR)
Real-time transcription with speaker diarization for meetings, interviews, and calls. Auto-summarized into action items.
AI Infographic Generator
Turn raw data, reports, or notes into clean visual infographics. Pick a style, get a polished export. All local.
How it works
From install to insights in three steps.
01
Install
Single installer for Windows or macOS. ~200 MB. No accounts, no signup, no telemetry.
02
Load models
Download an LLM, ASR, and OCR model from the in-app catalog — or drop in your own GGUF / ONNX files.
03
Work offline
Disconnect from the internet entirely. Vault AI keeps working — chat, RAG, OCR, transcription, all of it.
Privacy by architecture
Built for environments where the cloud is not an option.
Cloud AI tools require you to upload your data, your clients’ data, and your organization’s documents to someone else’s servers. For most regulated industries, that’s either a compliance violation or an unacceptable risk.
Vault AI inverts the model: every operation — embedding, inference, retrieval, transcription, OCR — happens on your local CPU or GPU. The application has no outbound network connections beyond optional update checks, which you can disable.
- No telemetry, no analytics, no “anonymous usage data”
- Encrypted document index using AES-256-GCM
- Air-gap install option for high-security environments
- Local audit log of all AI operations (Business tier)
- Open-weight models you can audit and replace
Pricing
Simple annual pricing.
One yearly subscription per user. All updates included. Cancel anytime.
Standard
For individuals
$
79
/year
Buy Standard
- Local LLM chat (7B–13B models)
- Document Q&A (RAG) with up to 5,000 documents
- 1 device activation
- Email support
- All updates included for 12 months
Most popular
Professional
For power users
$
149
/year
Buy Professional
- Everything in Standard
- Larger models (up to 70B with quantization)
- Unlimited documents in RAG index
- PII redaction & OCR
- Meeting notes with ASR
- AI infographic generator
- 2 device activations
Business
For teams & regulated industries
$
249
/year
Buy Business
- Everything in Professional
- Priority support
- 5 device activations
- Audit log export
- Air-gapped install support
- Volume licensing available
All tiers include a 30-day money-back guarantee. Volume licensing for 10+ seats: support@vault-ai.app
System requirements
What you need to run it.
Minimum
- CPU
- 4-core x86-64 (2018+) or Apple Silicon
- RAM
- 8 GB
- Storage
- 10 GB free (more for larger models)
- OS
- Windows 10/11, macOS 12+
Recommended
- CPU
- 8+ cores, AVX2 support
- RAM
- 32 GB+
- GPU
- NVIDIA 8 GB+ VRAM or Apple M-series
- Storage
- NVMe SSD with 50 GB+ free
FAQ
Common questions.
Does Vault AI really work with no internet?
Yes. After install and model download, you can disconnect entirely. The app has no required network calls — chat, RAG, OCR, ASR, and PII redaction all run locally. Update checks are optional and disabled in the Business tier’s air-gap mode.
Which AI models can I use?
Any GGUF-format LLM (Llama, Mistral, Qwen, Gemma, Phi, and more), any ONNX-format ASR/OCR/embedding model, and our curated catalog of pre-tested models with one-click download. You can also bring your own.
How does it compare to running Ollama or LM Studio myself?
Vault AI bundles inference (LLM/ASR/OCR), retrieval, document indexing, PII detection, transcription, and export into one workflow-focused app. You can absolutely build a similar stack yourself — Vault AI is for people who want it ready out of the box, with support, a license, and predictable behavior.
What about my data privacy?
Vault AI does not collect, transmit, or log any of your content. The license activation server records only the license key and a device fingerprint hash — never your documents, prompts, or model outputs. See our Privacy Policy for full details.
Can I use it for client work / regulated data?
That’s the primary use case. Many of our users are lawyers, doctors, accountants, and security analysts handling protected data. The Business tier includes air-gap support and audit logging suitable for HIPAA, GDPR, and similar compliance regimes — though final compliance certification depends on your overall workflow, not just the software.
Is there a refund policy?
Yes — 30-day money-back guarantee on all tiers. Email support@vault-ai.app with your license key.
What happens when my annual subscription expires?
Renew to keep using Vault AI and continue receiving updates. If you let your subscription lapse, the application will stop working until you renew. You can cancel auto-renewal at any time from the customer portal.
Stop sending your data to someone else’s computer.
Run AI the way it should be run — on your machine, on your terms.