id: 688
uid: hfDpC
insdate: 2026-03-10 19:05:10
title: Launch HN: RunAnywhere (YC W26) – Faster AI Inference on Apple Silicon
additional: 🚀 Launch HN: RunAnywhere (YC W26) – Faster AI Inference on Apple Silicon: RunAnywhere's MetalRT engine enables faster AI inference on Apple Silicon, outperforming existing solutions like llama.cpp and Apple's MLX, with significant speed boosts in LLM decode, speech-to-text, and text-to-speech, allowing for seamless on-device AI experiences without relying on cloud APIs. By leveraging custom Metal shaders and skipping framework overhead, MetalRT achieves remarkable performance gains, such as transcribing 70 seconds of audio in 101ms and synthesizing speech in 178ms, making it an ideal solution for developers seeking to build fast and efficient AI-powered applications on Apple devices.
category: Hacker News
md5:
guid: https://news.ycombinator.com/item?id=47326101
source_url: https://github.com/RunanywhereAI/rcli
updated:
image:
author_name: sanchitmonga22
author_link: