Model Library
Privacy-First Local AI Chat
PocketLlama is a fully local AI chat application for iOS and iPadOS. Run powerful LLMs like Gemma 3, Llama 3.2, Qwen 2.5, and Phi-3 directly on your device with no internet required.
Key Features
- 100% Local Inference — Your conversations never leave your device
- Curated Model Library — Choose from Gemma, Llama, Qwen, Phi and more
- Smart Download Manager — Real-time disk usage tracking
- Performance Metrics — Monitor tokens/second and memory usage
- Customizable Answers — Adjust response style and length
- Adaptive UI — Light/Dark modes with multi-device optimization
Technology Stack
- SwiftUI — Native iOS/iPadOS interface
- llama.cpp — Efficient local model inference
- GGUF Models — Optimized quantized model format
- Core Data — Persistent conversation storage
- Combine — Reactive data flow
Screenshots
A clean, modern interface with smooth animations and intuitive controls.
Storage & Models
Chat Interface
Settings
About This Project
PocketLlama was built to provide a private, offline AI assistant that respects user privacy. Unlike cloud-based AI services, all processing happens locally on your device. This means:
- No data is sent to external servers
- Works completely offline
- No subscription fees or API costs
- Full control over your conversations