PHPrivate AI (v1.63) - On-Device LLM Client
An Android application that enables running large language models (LLMs) locally on-device, ensuring complete data privacy and offline functionality without internet dependencies or data collection.Key Features
- Offline Processing: Full conversational AI capabilities without an internet connection.
- Privacy-Centric: Local execution ensures zero data collection or external logging.
- Chat History: Searchable and indexed conversation logs for local reference.
- Ad-Free Interface: Clean UI with no advertisements.
- Model Customization (Premium): Support for importing and running external GGUF model files.
Technical Specifications & Supported Models
| Model Name | Developer / Origin | Characteristics & Use Cases |
|---|---|---|
| Gemma 2 (Default) | Lightweight, state-of-the-art model optimized for general text generation tasks. | |
| Llama 3.2 | Meta | Advanced model optimized for diverse language processing tasks. |
| Phi-3.5 mini | Microsoft | Lightweight footprint featuring strong reasoning capabilities. |
| Orca mini | Independent | General-purpose model optimized for entry-level hardware and low-RAM devices. |
| Rocket 3B | Independent | Compact 3-billion parameter model with high benchmark efficiency. |
| Mistral | Mistral AI | High-performance open-source foundational model. |
| NeuralBeagle | Independent | Fine-tuned model excelling in instruction following and complex reasoning. |
| OpenHermes | Independent | Versatile assistant fine-tuned with enhanced code generation capabilities. |
| WizardLM | WizardLM Team | Advanced language model trained without alignment constraints. |
| Silicon Maid | Independent | Specialized model fine-tuned for creative writing and role-playing tasks. |
Application Version Details
- App Version: v1.63
- Target Audience Age Rating: 12+
- Format Support: GGUF (Premium tier required for custom model sideloading)