The 57B is a mixture of experts, with only 14B parameters “active” at one time. So it has the memory footprint of a 57B but the compute footprint of a 14B. It

The 57B is a mixture of experts, with only 14B parameters “active” at one time. So it has the memory footprint of a 57B but the compute footprint of a 14B. It won’t fit on consumer GPUs but it will fit on Apple Silicon if there’s enough system RAM, and will be pretty fast (like Mixtral 8x7B).