Provides a unified local API for ASR, TTS, translation, and image generation optimized for Apple Silicon.
MLX Local Serving (MLS) is a high-performance, on-device AI toolkit designed specifically for Mac hardware using the MLX framework. It enables Claude to interact with a suite of resident models including Qwen3-ASR for speech recognition, Qwen3-TTS for customizable voice synthesis, TranslateGemma for 55-language translation, and Z-Image-Turbo for rapid image generation. By keeping models resident in GPU memory, the skill ensures minimal latency and maximum privacy for multimedia and linguistic tasks without reliance on external cloud APIs.
Key Features
01Hardware-accelerated performance optimized for Apple Silicon M-series chips
02Unified local API for ASR, TTS, Translation, and Image Generation
03High-speed batch processing for text files and multi-sentence translation
0410 GitHub stars
05Resident GPU memory management to eliminate model cold-start delays
06Customizable 'VoiceDesign' TTS with natural language character instructions
Use Cases
01Building privacy-first voice assistants and transcription tools on local hardware
02Rapid prototyping of visual assets and UI designs using on-device image generation
03Automating multilingual content workflows with localized translation and speech
