EdgeAI: Local AI Gallery

About this app

<p>EdgeAI runs large language models directly on your iPhone—so your chats and photo questions stay on your device, not in the cloud.</p>
<p>PRIVATE, OFFLINE AI<br />
• No sign-in required to start chatting<br />
• Inference runs locally after you download a model<br />
• Use on planes, poor networks, or anywhere you want more privacy<br />
• Your prompts and images are not sent to our servers for AI processing</p>
<p>CHAT WITH OPEN MODELS<br />
• Built-in model library with size guidance for your device<br />
• Pick lightweight models for everyday questions or larger models when you have more RAM<br />
• Supports popular on-device families including Qwen and Gemma<br />
• Streaming replies for a natural chat experience</p>
<p>VISION & MULTIMODAL<br />
• Ask questions about photos you choose from your library<br />
• Multimodal models for image understanding on supported devices<br />
• Great for quick captions, explanations, and visual Q&A—without uploading pictures to the web</p>
<p>BUILT FOR REAL PHONES<br />
• Recommendations based on available device memory<br />
• Download only the models you want<br />
• Optimized native runtimes on iOS (MLX, LiteRT-LM, and llama.cpp where supported)</p>
<p>WHO IT’S FOR<br />
• Students, travelers, and professionals who want offline AI<br />
• Privacy-conscious users who prefer on-device assistants<br />
• Developers and enthusiasts exploring local LLMs on iPhone and iPad</p>
<p>HOW IT WORKS<br />
1. Open the model library and download a model (one-time, over Wi‑Fi recommended).<br />
2. Start a chat—or attach an image with a vision-capable model.<br />
3. Chat offline anytime after the model is installed.</p>
<p>IMPORTANT NOTES<br />
• Model files are large; downloads use storage and are best on Wi‑Fi.<br />
• Performance depends on your iPhone model and free RAM; heavier models need newer, higher-memory devices.<br />
• EdgeAI is a client for on-device open models; it is not affiliated with Google, Alibaba, or other model publishers.</p>
<p>Questions or feedback? Contact us at [email protected]</p>