Home › GPUs & Macs › Apple M4 Max (40-core GPU) vs NVIDIA RTX 4090
Apple M4 Max (40-core GPU) vs NVIDIA RTX 4090 for local LLMs
Across 4 tests measured on both, the NVIDIA RTX 4090 is faster in 4 and generates a median 1.83× the tokens/s of the Apple M4 Max (40-core GPU).
On Llama 2 7B (Q4_0, llama-bench) the Apple M4 Max (40-core GPU) generates 83.1 tokens/s and the NVIDIA RTX 4090 190 tokens/s (measured).
Same model, same quant, same test
| Model | Quant / test | M4 Max (40-core GPU) tok/s | RTX 4090 tok/s | Faster | Sources and run details |
|---|---|---|---|---|---|
| Llama 2 7B | Q4_0 llama-bench | 83.1 Metal | 190 Vulkan | RTX 4090 +129% | llama.cpp #4167 M4 Max (40-core GPU) · llama.cpp #10879 RTX 4090 |
| Llama 3.1 8B Instruct | Q4_K_M LocalScore | 52.9 llamafile | 105 llamafile | RTX 4090 +99% | LocalScore M4 Max (40-core GPU) · LocalScore RTX 4090 |
| Llama 3.2 1B Instruct | Q4_K_M LocalScore | 180 llamafile | 292 llamafile | RTX 4090 +62% | LocalScore M4 Max (40-core GPU) · LocalScore RTX 4090 |
| Qwen2.5 14B Instruct | Q4_K_M LocalScore | 28.9 llamafile | 49.1 llamafile | RTX 4090 +70% | LocalScore M4 Max (40-core GPU) · LocalScore RTX 4090 |
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) | NVIDIA RTX 4090 | |
|---|---|---|
| Memory | Unified memory | 24 GB |
| Memory bandwidth | 546 GB/s | 1008 GB/s |
All results for the Apple M4 Max (40-core GPU) · All results for the NVIDIA RTX 4090 · More comparisons