Nyth AI brings high-performance, on-device intelligence to your iPhone. Unlike standard AI apps that rely on cloud servers, Nyth AI runs models locally using compiled Metal kernels via MLC-LLM and Apache TVM. This ensures 100% privacy and full functionality in airplane mode with zero latency from data transfers. Features: * Models: Qwen 2.5, Llama 3.2, and Gemma 2. * No accounts, subscriptions, or data logging. * Optimized GPU memory management for mobile stability.
No reviews yetBe the first to leave a review for Nyth AI
Maker
📌
Hi Product Hunt! 👋
I’m the creator behind codelynther, and I’m excited to launch Nyth AI today.
Most AI apps are just wrappers for cloud APIs, which means your data is constantly leaving your device. I built Nyth AI to change that by bringing the actual intelligence to the silicon. By leveraging MLC-LLM and Apache TVM, I’ve optimized the inference path to run 1B and 2B parameter models natively on iOS hardware.
Why Nyth AI?
* 100% Offline: It works in tunnels, on planes, and in dead zones because the "brain" is on your phone.
* Absolute Privacy: Your prompts and outputs never leave your device. No accounts, no data logging, no cloud.
* Performance Optimized: Custom GPU memory management and prefill chunking ensure stability even during heavy generation tasks.
Whether you need a fast response from Flash (Qwen 2.5) or more complex logic from Sage (Gemma 2), Nyth AI provides a smooth, private experience.
I’d love to hear your thoughts on the speed and responsiveness!