News for hackers. Stuff that matters.
Re:Show HN: We made our own inference engine for Apple Silicon by zdw on Tuesday July 15, 2025 @12:16PM (#44572724)
How does this bench compared to MLX?
Re:Show HN: We made our own inference engine for Apple Silicon by jasonjmcghee on Tuesday July 15, 2025 @12:41PM (#44573093)
I use MLX in lmstudio and it doesn't have whatever issues llama cpp is showing here.

Qwen3-0.6B at 5 t/s doesn't make any sense. Something is clearly wrong for that specific model.