锦囊AI

v1.4 runs 7B/9B-class models locally(tested on iPhone15 Pro Max 8GB Ram), with LaTeX rendering and adaptive memory.

JinNang runs entirely on your iPhone — your conversations never leave the device. No account, zero data collection. • On-device inference: chat with no network connection (internet is only needed to download models) • A lightweight model is built in; download 7B / 9B-class models for more capability • 11 device tools: calculator, date/time, files, system, battery, translate, Safari, clipboard, reminders, location, contacts • LaTeX math rendering, knowledge base Q&A, and Gomoku / Chess with AI commentary • Tunable context window, temperature, thinking budget, and memory profile • Sampling follows each model's official recommendation (Qwen3 / Qwen3.5 / DeepSeek-R1 / VibeThinker) • Privacy by design: no tracking, no data collection — your data stays on your device • Requirements: tested on iPhone 15 Pro Max (8GB RAM); 12GB+ devices have more headroom for large models. • Pro unlocks 4B / 7B / 9B-class models, the chess coach, parameter tuning, and custom prompts.