Built for work that can't touch the cloud

Private AI for lawyers, doctors, therapists, and accountants.

LocalChat is an on-device LLM that runs on your Mac. It reads the whole file. Nothing leaves your laptop.

LocalChat app on macOS with a private AI workspace: recent conversations in the sidebar, a message box with file upload and audio recording, and an on-device Gemma model selected

Why local matters

Cloud AI breaks rules you've already agreed to follow

You work under rules. Privilege. HIPAA. Your licensing board. ChatGPT respects none of them — so your real work never gets near it.

Privileged work shouldn't sit on someone else's server

Paste a case file into ChatGPT and it lives on a server you don't own. Their staff can read it. Their next model can train on it. Their lawyers can hand it over.

Patient details don't belong in a chat tool

ChatGPT, Claude, and Gemini weren't built to hold patient information. One pasted note can put you on the wrong side of HIPAA.

Client financials don't belong in cloud AI

Tax returns, payroll exports, P&Ls, and engagement letters carry bank details, SSNs, and private business numbers. Accountants need answers without creating another copy online.

Therapy notes don't belong in training data

Even with the opt-out checked, your intake forms and session notes still go to their servers. Whatever you upload, assume someone else has a copy.

LocalChat is the local LLM that keeps your work on your Mac

The ChatGPT alternative that runs on your laptop. Your files, your questions, and the answers stay on the same machine. No server to leak from. No account to log into. Nothing for anyone to hand over.

Built for

The people who can't put their work in the cloud

Lawyers
Read contracts, depositions, and discovery on your Mac. Ask clause-level questions and get page citations — without uploading a single file.
Therapists
Summarize session notes and intake forms on-device. HIPAA compliance stays simple because the file never leaves your laptop.
Accountants
AI for accountants in busy season: skim tax returns, P&Ls, and engagement letters. Client numbers never touch a cloud account.
Consultants
Brief yourself on client docs, draft proposals, polish decks. Nothing to clear with anyone's IT team.
Writers & researchers
Hand it a whole manuscript or stack of sources. Get outlines, edits, and citations — your unpublished work stays unpublished.
Travellers
Real AI on a plane, in a basement hotel, on patchy hotel Wi-Fi. No login screen between you and the answer.

Model library

A curated catalog, picked for Mac

Reasoning models, vision models, embeddings for memory, even an image generator — every one tested on Apple Silicon and one click to download. Then it works without an internet connection, ever.

Qwen3.5 9B

Vision

Alibaba9B · 5.7 GB

GLM 4.6V Flash

Vision

Z.ai~9B · 6.2 GB

DeepSeek R1 (Qwen3 8B)

Reasoning

DeepSeek8B · 5 GB

Qwen3-VL 30B A3B

Vision

Alibaba30B · 19 GB

Gemma 4 E4B

Vision

Google DeepMind4B · 5 GB

gpt-oss 20B

Reasoning

OpenAI20B · 12 GB

Qwen2.5 14B Instruct

Alibaba14B · 8.5 GB

Granite 4.0 Micro

IBM3B · 2.1 GB

olmOCR 2

OCR

Allen AI7B · 5 GB

Z-Image Turbo

Image gen

Tongyi Lab5 GB

Qwen3 Embedding

Embedding

Alibaba0.6B · 0.6 GB

Gemma 3

Google DeepMind1B · 0.3 GB

Plus any GGUF model on Hugging Face

Paste a URL or drop in a .gguf file — LocalChat handles the rest.

qwen · deepseek · gemma · glm · granite · gpt-oss · olmocr · …

Features

A private LLM you can hand the real file to

Built for the work you do — reading long files, comparing answers, drafting. With none of the redacting before you can ask a question.

Drop in the whole file

Hand it a 200-page deposition, a full patient chart, a contract, an intake packet. PDFs, markdown, CSVs, spreadsheets — LocalChat reads the whole thing on your Mac, no copy-paste.

PDFMDCSVXLSXPNGTXT

No account, no logins, no usage tracking

There's no LocalChat account to sign up for. Your prompts, files, and answers stay on your laptop — there's nothing for anyone else to look at.

A project workspace per matter

Each project keeps its own files, conversations, and embedded memory. Toggle 'answer only from this project's files' and the model stays inside the case.

Plug in your Obsidian vault

Point LocalChat at a vault and it scans the notes, follows links, and turns them into searchable memory. Your second brain becomes the assistant's first source.

On-device AI, on a plane or anywhere

Once it's installed, no internet required. Keep working in places where ChatGPT and Claude are blocked or out of reach.

Pick the right LLM for the job

A curated catalog of 25+ on-device models — reasoning, vision, OCR, image gen, embeddings. Swap mid-thread; the conversation stays.

+18

Local by design

Work with AI where your private files already live

Sensitive notes, client documents, internal research, and personal drafts stay on your Mac. LocalChat is a private AI assistant without the upload step.

Messages processed on-device: 100%
Cloud accounts required: 0
Open-source models supported: 300+
Offline availability: Always

Roadmap

More Mac-native context, same private foundation

Buy once and keep receiving product updates. The roadmap expands what LocalChat can read and create without changing where your data lives.

In development

Apple Notes and projects

Index notes, folders, and project files as private context.

Coming soon

Voice input

Dictate prompts while LocalChat transcribes on your Mac.

In development

iMessage search

Ask natural-language questions across local message history.

Coming soon

Image generation

Generate visuals with local diffusion models.

Planned

Private web search

Give local models fresh context without ad-tech tracking.

Planned

MCP plugins

Connect GitHub, Notion, Slack, and internal tools.

Trusted workflows

Built for work people do not want to upload

"I can review tax packets, engagement letters, and client spreadsheets without pasting private numbers into a cloud chat."

Priya Nair

Accountant

"For contracts and discovery, I need the whole file in context. LocalChat lets me ask questions without uploading privileged material."

Daniel Brooks

Lawyer

"Patient notes and intake PDFs stay on my Mac. That makes summarizing charts useful without creating another data exposure."

Maya Patel

Doctor

"Session notes and treatment plans stay in my workspace. I can draft follow-ups without sending private notes to a server."

Lena Ortiz

Therapist

100% on-device processing

300+ open-source models

No account required

Pricing

Own the app. Stop renting AI.

One payment for a Mac-native AI workspace. No subscription just to keep talking to your own files.

50%off

Launch deal for the first 100 customers

74 left

Single

For one Mac and one private AI workspace.

$49.50$99

All local chat features
300+ model support
Document chat
1 year of updates
Email support

Buy license

Family

Popular

For up to 5 devices at home or across a small team.

$199.50$399

Everything in Single
Up to 5 devices
Priority support
Shared purchase management
1 year of updates

Buy license

Team

For organizations that need deployment and volume terms.

Custom

Volume licensing
Deployment guidance
Custom procurement
Dedicated support
Security review help

Contact sales

30-day money-back guarantee

Instant download after purchase

Runs 100% on your device

FAQ

Questions before you go local

Your files. Your Mac. Nothing else.

Stop pasting one paragraph at a time.

Hand it the full file. It reads the whole thing on your Mac. It tells you where each answer came from. Nothing leaves your laptop.

A private AI chatbot that runs on your Mac
Reads whole files, not pasted snippets
Nothing leaves your Mac
One-time license, no AI subscription

Download for Mac View pricing