u/DankpawsDev@DankpawsDev · Oct 6Made a custom MLX quant sized for a 96GB machine after finding existing models lacked the right balance of speed and headroom.reddit.com