3 ms·

How does this compare to llama.cpp in terms of performance?

• solarkraft 6mo ago
  MLX is a bit faster (low double digit percentage), but uses a bit more RAM. Worthwhile tradeoff for many.
  • ysleepy 6mo ago
    On my M4 Pro MLX has almost 2x tok/s