The Open-Source AI Stack
RSS
All hardware

Hardware · Apple unified

Apple MacBook Pro (M4 Max)

Apple

Bandwidth is binned by GPU core count: 410 GB/s on the 32-core part and 546 GB/s on the 40-core part. The 40-core figure is used here.

CPU · GPU · NPU 546 GB/s memory bus 128 GB Unified memory
Apple unified. Bus width tracks bandwidth (546 GB/s, sets decode speed); the box tracks capacity (128 GB, sets what fits). Unified memory is one pool shared by CPU and accelerator.

Specs

Memory
128 GB Unified LPDDR5X
Bandwidth
546 GB/s
Power
40 W
Form factor
soc
Interconnect
none
Released
2024-10

What it runs (single unit, Q4_K_M, 4K context)

Model Fits? Decode ceiling
Llama 3.1 8B Instruct yes ~108 tok/s
Llama 3.3 70B Instruct yes ~13 tok/s
Qwen 2.5 72B Instruct yes ~13 tok/s
DeepSeek-V3 no —

Ceiling is the theoretical rooflineruntimeA performance model that bounds throughput by either compute or memory bandwidth, whichever is the limiting resource for an operation's arithmetic intensity. Open full entry ; open the explorer to set quant, context, and runtime and see the realistic range.

Sources