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

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.