LMBox - Private Offline AI

LMBox - Private Offline AI

Voice & PDF chat, on-device

0 ratings
Free

Rating summary

Details

Released
October 16, 2026
Updated
October 25, 2026
LMBox - Private Offline AI screenshot #1 for iPhone
LMBox - Private Offline AI screenshot #2 for iPhone
LMBox - Private Offline AI screenshot #3 for iPhone
LMBox - Private Offline AI screenshot #4 for iPhone
iphone
ipad
🖼️Get Icon
Icons↘︎

About

Your private AI assistant. All data processed on-device. Chat, ask questions, write, and explore ideas on your iPhone or iPad. LMBox runs language models directly on your device. AI chat, recording transcription, text recognition, and PDF Q&A are processed locally, without an account or sending your conversations, recordings, photos, or files to cloud AI. Download the required models first, then keep working offline. ON-DEVICE AI CHAT Choose a model that fits your device for text Q&A, drafting, and brainstorming. Stream replies, stop generation, and save or search your conversation history. TRANSCRIBE RECORDINGS, THEN ASK Record, pause, save, preview, and replay audio. Transcribe recordings on-device, keep the text with your conversation, ask follow-up questions, or run recognition again. ASK ABOUT PDFs AND SCANNED TEXT Add PDFs, scanned PDFs, photos, and text files. View original attachments, processing progress, and the full extracted text. Find relevant passages and PDF page numbers based on your questions. MANAGE YOUR PERSONAL CONTENT View local storage and separately remove personal content, caches, chat models, or speech recognition resources. No account, ads, or third-party product analytics. BEFORE YOU START The interface supports English, Simplified Chinese, and Traditional Chinese. Switch the app language in system settings; existing conversations and original files remain unchanged. Chat models and the initial speech recognition resources of about 1 GB require an internet connection and sufficient storage to download. Once the required models are ready, processing works offline. Performance and available models depend on your device resources. Photos support text recognition only, not visual scene understanding. Each PDF supports up to 200 pages or 1 million characters, with the extracted range shown in the app. Questions about long material remain subject to model context limits. Recognition and AI responses can be incorrect; check the original sources. Personal content uses system file protection. System backups depend on your device settings. On-device processing does not imply additional app-level end-to-end encryption.
Show more

What's New in LMBox

1.0.1

October 25, 2026

Updated the App Store description and feature information to clarify on-device AI chat, transcription, and PDF Q&A.

Developer apps