PortableLM is a fully air-gapped, zero-dependency, plug-and-play local AI environment designed to run seamlessly from your local hard drive or a portable USB/SSD. It bypasses complex installations — natively executing large language models, image generation, and high-quality text-to-speech directly on your hardware with no internet required.

With a unified architecture, you can initialize your AI models once and carry them with you across Windows, macOS, Linux, and Android (Termux).

• Multi-Modal AI Hub: A single interface for Text Chat (LLM), Image Generation (Stable Diffusion), and Text-to-Speech (Piper TTS).
• Persistent Image Library: Generated images are automatically saved to Shared/chat_data/generated_images/ with sidecar metadata. Browse, view full-screen, reuse prompt parameters, download, or delete images directly from the built-in Image Library gallery.
• Aspect Ratio & Live Step Progress: Image generator includes Aspect Ratio selectors (16:9 Landscape, 9:16 Portrait, 1:1 Square, Custom) and a real-time progress bar showing active step counts (Step 1 / 20 ... 20 / 20) and ETA calculations.
• Persistent Audio Library: Generated TTS speech files are saved to Shared/chat_data/tts_output/ with sidecar metadata. Listen with the inline player, reuse text, download, or delete clips from the Audio Library gallery.
• Lightweight Model Options: Includes ultra-fast local LLM models like Liquid AI LFM2.5 230M (LFM2.5-230M.Q4_K_M.gguf), Gemma 2, Llama 3, and Qwen.
• Zero Dependency Setup: Ships with portable Python and isolated engine binaries. No system permissions, registry edits, or package managers required.
• Cross-Platform: Uses an intelligent Shared volume system — download your AI models once, and use them natively on Windows, macOS, Linux, and Android (Termux).
• Fully Offline: Runs completely air-gapped after initial setup. Your data never leaves your machine.
• Network Proxied UI: The custom Python HTTP server serves a blazing-fast dark mode UI. Access the AI from your phone or tablet on the same WiFi — no CORS headaches.

Powered by Ollama, run world-class models like Liquid AI LFM2.5 230M, Gemma 2, Llama 3, and Qwen entirely locally. Full support for custom .gguf models and advanced system instructions.

Powered by Stable Diffusion (stable-diffusion.cpp), generate high-quality images using CyberRealistic v3.3. Supports aspect ratio presets (16:9, 9:16, 1:1, Custom), sampler selection, seeds, and real-time step progress tracking.

Powered by Piper, transform text into natural-sounding speech instantly. Includes 5+ high-quality female and male voices (Amy, Lili, Kusal, Arctic, Lessac, Alan) with persistent sidecar storage and full Audio Library management.

• Storage: USB 3.0+ flash drive or SSD with at least 12 GB free (for Chat + Image + TTS models).
• RAM: At least 4 GB - 8 GB for base models (e.g. Liquid AI LFM2.5 230M), 16 GB recommended for multi-modal workloads.
• OS: Windows 10/11, macOS (Intel/Silicon), modern Linux distributions, or Android (Termux).

[PortableLM Drive]
 ├── 📁 Android    # Native Termux / Android installers & scripts
 ├── 📁 Linux      # Native Linux (Ubuntu/Debian) launchers & installers
 ├── 📁 Mac        # Native macOS (Intel/Silicon) launchers & installers
 ├── 📁 Windows    # Native Windows launchers & installers
 └── 📁 Shared     # Unified Cross-Platform Data System
      ├── 📁 bin         (Isolated engine binaries: Ollama, Stable Diffusion, Piper)
      ├── 📁 chat_data   (Persistent chat history, generated_images, and tts_output)
      ├── 📁 models      (LLM weights, SD checkpoints, and Piper voice models)
      └── 📁 vendor      (Local UI assets: JS/CSS/Fonts for 100% offline usage)

Run the install script for your platform. This will prompt you to select your preferred LLM model (including Liquid AI LFM2.5 230M) and download execution engines.

The server will start, and your browser will open to http://localhost:3333.

Use your PC's AI from your phone or tablet on the same network:

1. Ensure the app is running on your PC.
2. The terminal will show a Network Access IP (e.g., http://192.168.1.15:3333).
3. Open that URL on your mobile browser.
4. Generate text, images, or speech directly from your mobile device!

PortableLM runs 100% offline after initial setup. No data—including chat history, generated images, or voice outputs—ever leaves your machine. No telemetry, no analytics, no cloud dependencies, no hidden phone-home requests.

What we do NOT collect:

• Personal information
• Usage statistics
• AI prompts or responses
• Generated content
• Hardware or system data

By using PortableLM, you retain full ownership of all content you generate.

This project is licensed under the MIT License - see the LICENSE file for details.

Copyright (c) 2026 Mohd Abuzar

PortableLM — Your Personal, Portable AI Command Center.