Personal GPT

Functions & Features

• Completely Offline Operation: No internet required, data never leaves the device, ensuring ultimate privacy with No Cloud, No Tracking, and No Logins.
• Seamless System Integration: Deeply integrated with Siri and Apple Shortcuts to create powerful AI-driven workflows without writing code; supports the x-callback-url specification to seamlessly add on-device AI functionality to over 70 popular iOS/macOS apps.
• macOS System-Wide Services: Provides AI-powered text enhancement tools across all macOS apps, supporting grammar correction, summarization, and multiple languages including English and select Western European languages.

Models & Technology

• Extensive Open-Source Model Support: Features leading open-source models like DeepSeek R1 Distill, Llama 3.3, Phi-4, Qwen3, and Google Gemma 2. Offers quantized versions ranging from 1B to 70B parameters, tailored for devices with 4GB to 64GB+ RAM.
• OmniQuant Quantization Algorithm: Utilizes the state-of-the-art OmniQuant quantization technique. It handles weight outlier values using an optimization-based learnable weight clipping mechanism, preserving the model's weight distribution with exceptional precision compared to standard RTN quantization. Paired with optimized model-specific Metal kernels, it delivers fast, high-quality on-device text generation.

Target Users & Use Cases

Ideal for individual users who highly value data privacy, as well as for scenarios requiring AI assistance for text generation, language help, content summarization, and automated workflows in environments without internet access (e.g., during flights).

Pricing Model

Available via a one-time purchase with no subscription fees. A single purchase unlocks the app across iPhone, iPad, and Mac platforms and enables Family Sharing for up to six relatives, priced at around $10 USD. Bootstrapped by an independent team of two developers with no venture capital.