Home › GPUs & Macs › Apple M4 Max (40-core GPU) vs Apple M5 Max (32-core GPU)
Apple M4 Max (40-core GPU) vs Apple M5 Max (32-core GPU) for local LLMs
Across 3 tests measured on both, the Apple M5 Max (32-core GPU) is faster in 3 and generates a median 1.18× the tokens/s of the Apple M4 Max (40-core GPU).
On Llama 3.1 8B Instruct (Q4_K_M, LocalScore) the Apple M4 Max (40-core GPU) generates 52.9 tokens/s and the Apple M5 Max (32-core GPU) 62.2 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | M4 Max (40-core GPU) tok/s | M5 Max (32-core GPU) tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Llama 3.1 8B Instruct | Q4_K_M LocalScore | 52.9 llamafile | 62.2 llamafile | M5 Max (32-core GPU) +18% | LocalScore M4 Max (40-core GPU) · LocalScore M5 Max (32-core GPU) |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 180 llamafile | 229 llamafile | M5 Max (32-core GPU) +27% | LocalScore M4 Max (40-core GPU) · LocalScore M5 Max (32-core GPU) |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 28.9 llamafile | 34.1 llamafile | M5 Max (32-core GPU) +18% | LocalScore M4 Max (40-core GPU) · LocalScore M5 Max (32-core GPU) |
Generation speed in tokens/s, best measured backend for each chip (shown under the number). Backends can differ between the two chips; check them before drawing conclusions. Measured results only, nothing estimated.
Specs
| Apple M4 Max (40-core GPU) | Apple M5 Max (32-core GPU) | |
|---|---|---|
| Memory | Unified memory | Unified memory |
| Memory bandwidth | 546 GB/s | 460 GB/s |
All results for the Apple M4 Max (40-core GPU) · All results for the Apple M5 Max (32-core GPU) · More comparisons