One-liner, as posted
Faster AI Inference on Apple Silicon
36 characters (fits the 50-character limit).
Excerpt
We built a fast inference engine for Apple Silicon. LLMs, speech-to-text, text-to-speech – MetalRT beats llama.cpp, Apple's MLX, Ollama, and sherpa-onnx on every modality we tested. Custom Metal shaders, no framework overhead.
Short excerpt from the founders’ Launch HN post. Read the full original at the link below.
Read the launch post on Hacker News Company website
Posted 2026-03-10 · 240 points · 153 comments. Batch as stated by the founders in the post title.
Working on your own?
Get line-by-line feedback on your one-liner and application answers.
Review my application