LocalChat Documentation

LocalChat is a local LLM app for macOS. It runs AI models entirely on your Mac, so your conversations stay private - no data is sent to external servers.

It's built for work that can't touch the cloud: contracts, patient notes, client financials, session notes, unpublished drafts. If your job involves files you'd never paste into a cloud chatbot, LocalChat gives you the same assistant workflow without the upload.

What LocalChat does

LocalChat lets you chat with AI models the same way you'd use ChatGPT or Claude, but with one key difference: the AI runs on your computer, not in the cloud. This means:

  • Complete privacy - Your conversations never leave your Mac
  • Works offline - Use AI without an internet connection
  • No subscriptions - One-time purchase, use forever
  • No rate limits - Chat as much as you want

How it works

  1. You download LocalChat and install it on your Mac
  2. During setup, you choose an AI model to download
  3. The model runs locally using your Mac's processor and memory
  4. You chat with the AI just like any other chat app

The AI model files are stored on your Mac, typically in the range of 2-8GB depending on the model size. Larger models generally produce better responses but require more memory.

Supported models

LocalChat ships with a curated catalog of 25+ models, tested and picked for Apple Silicon:

  • Qwen - Multilingual models with vision and embedding support
  • DeepSeek - Reasoning-focused models
  • Gemma - Google DeepMind's lightweight, efficient models
  • GLM - Vision models from Z.ai
  • gpt-oss - OpenAI's open-weight reasoning models
  • olmOCR - Allen AI's document OCR model

Beyond the curated catalog, you can also paste a Hugging Face URL or drop in any .gguf file to run it.

You can download multiple models and switch between them depending on your task.

System requirements

RequirementMinimumRecommended
macOS13.0 (Ventura)Latest version
ProcessorAnyApple Silicon (M1/M2/M3/M4)
RAM8GB16GB+
Storage10GB free50GB+ free

Apple Silicon Macs get the best performance because LocalChat can use the GPU for faster responses.

Key features

Model management

Browse, download, and organize AI models from the built-in model gallery. No command line needed. See Models for details.

Image analysis

Use vision-enabled models to analyze screenshots, photos, and diagrams. Extract information, describe content, or get explanations. See Image analysis.

Document processing

Drag and drop PDFs, spreadsheets, or text files to analyze them with AI. LocalChat reads the whole file - summarize it, ask questions, or extract data, all locally. See Chat with Documents.

Projects

Group related files and conversations into workspaces. Each project keeps its own documents and memory, and you can restrict answers to only the project's files. See Projects.

Obsidian vault

Connect an Obsidian vault and LocalChat indexes your notes, follows the links between them, and makes years of notes searchable with AI. See Obsidian vault.

Screenshot to code

Convert website designs and UI mockups into working HTML and CSS code using vision models. See Screenshot to code.

Offline mode

After downloading your models, you can use LocalChat without any internet connection. Great for travel, privacy-sensitive work, or unreliable connections.

Conversation history

Your chat history is stored locally on your Mac. Search previous conversations and pick up where you left off.

Getting started

Ready to try LocalChat? Follow these guides: