LocalChat.app

About LocalChat.app

LocalChat is a Mac application that runs large language models entirely on-device, without sending data to a cloud server. It reads full documents directly — including PDF, Markdown, CSV, XLSX, PNG, and TXT files — rather than requiring text to be pasted in, and organizes work into per-project workspaces that can be restricted to answer only from files within that project. It also supports connecting to an Obsidian vault, indexing notes and links as searchable memory.

The application ships with a curated library of over 300 open-source models optimized for Apple Silicon, spanning reasoning, vision, OCR, image generation, and embedding models, and supports loading any GGUF model from Hugging Face by URL or file. Models can be swapped mid-conversation. No account or login is required, and no internet connection is needed once the app is installed and a model is downloaded.

It targets professionals handling sensitive or regulated material — lawyers, therapists, doctors, accountants, consultants, and researchers — where cloud-based AI tools may conflict with confidentiality, privilege, or compliance obligations. Pricing is a one-time license rather than a subscription: a Single license (one Mac) is listed at $49.50 (discounted from $99), a Family license (up to 5 devices) at $199.50 (discounted from $399), both including one year of updates, and a custom-priced Team license for organizations needing volume licensing and deployment support. Purchases include a 30-day money-back guarantee.

Key features

• Runs on-device on a Mac with no internet connection required after setup
• Reads whole files directly — PDF, Markdown, CSV, XLSX, PNG, TXT — no copy-paste
• Curated catalog of 25+ on-device models plus any GGUF model from Hugging Face
• Project workspaces that scope answers to only that project's files
• Indexes an Obsidian vault as searchable memory
• No account, login, or usage tracking — nothing leaves the laptop

Frequently asked questions