qwen3

A native macOS app that allows users to chat with a local LLM that can respond with information from files, folders and websites on your Mac without installing any other software. Powered by llama.cpp.

iOS voice note app with local Whisper.cpp voice transcription & Qwen 3 local LLM

On-device speech recognition for Apple Silicon, powered by MLX.

Native Swift + Metal local LLM inference for Qwen 3.6 35B-A3B on Apple Silicon. Streams active MoE experts from SSD for ~5 GiB Metal allocation; reads MLX Q4 Safetensors and includes a macOS chat app and benchmarks.

Local TTS Package for macOS w/ Cloning & Batch Generation. iOS Soon™

Local menu-bar dictation for macOS, built for mixed Chinese-English speech. Qwen3-ASR on MLX — audio never leaves your Mac.

Nocturne: A 100% local, private voice dictation client for macOS inspired by Wispr Flow. Transcribes using Parakeet (CoreML/Neural Engine) and polishes text with local AI (internal self-managed Ollama or Apple Intelligence). Fully offline, featuring usage stats, custom dictionary, snippets, and native support for 12 languages

Reproducible Qwen3-1.7B Core AI INT4 GPU conversion, physical-device benchmarks, and W8/ANE comparison.

On-device, commercial-safe text-to-speech for macOS & iOS. One TTSEngine protocol over Apple system voices, Kokoro & PocketTTS (CoreML/ANE), Chatterbox & Qwen3-TTS (MLX/GPU), and Supertonic (ONNX, 31 languages).

Pipeline RAG 100% on-device para iOS: recuperação semântica com NaturalLanguage embeddings e geração de texto local via Qwen2.5-3B rodando com MLX sem enviar dados para servidores externos.