M4 Max (40-core GPU) — benchmark record

M4 Max (40-core GPU) local LLM benchmarks across 48 GB, 64 GB, 128 GB RAM tiers on Apple Silicon Mac. 20 published rows across 6 models with explicit evidence state and RAM-tier comparison. Peak published speed is 182.6 tok/s.

20Benchmark rows

6Models tested

3RAM configurations

182.6Fastest avg tok/s

RAM configurations

Each configuration differs only in unified memory. More RAM = larger models fit. Throughput is similar across RAM tiers at the same model size.

48 GBM4 Max (40-core GPU, 48 GB)3 rows · 3 models179.0 tok/s peak 64 GBM4 Max (40-core GPU, 64 GB)14 rows · 6 modelsLab ✓180.2 tok/s peak 128 GBM4 Max (40-core GPU, 128 GB)3 rows · 3 models182.6 tok/s peak

All benchmark rows — M4 Max (40-core GPU)

Sorted by avg tok/s descending. Click source badge to see original measurement.

Chip (RAM)	Model	Quant	RAM req.	Avg tok/s	Prompt tok/s	Runtime	Source
M4 Max (40-core GPU, 128 GB)	Llama 3.2 1B	Q4_K - Medium	—	182.6 tok/s	3833.8 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Llama 3.2 1B	Q4_K - Medium	—	180.2 tok/s	3857.5 tok/s	llamafile	ref
M4 Max (40-core GPU, 48 GB)	Llama 3.2 1B	Q4_K - Medium	—	179.0 tok/s	3750.6 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q4_G32	2.8 GB	149.1 tok/s	2838.4 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q4	2.5 GB	148.1 tok/s	2976.7 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q5	3.3 GB	143.2 tok/s	2736.2 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q5_G32	3.5 GB	143.0 tok/s	2754.5 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q6	4.0 GB	136.6 tok/s	2735.7 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 4B	Q8	5.1 GB	111.5 tok/s	1780.6 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 30B-A3B	Q4	16.1 GB	92.1 tok/s	822.6 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 30B-A3B	Q5	18.1 GB	84.9 tok/s	819.8 tok/s	MLX	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 30B-A3B	Q6	21.9 GB	76.7 tok/s	817.6 tok/s	MLX	ref
M4 Max (40-core GPU, 48 GB)	Llama 3.1 8B	Q4_K - Medium	—	55.1 tok/s	663.4 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 30B-A3B	Q8	29.8 GB	52.6 tok/s	772.6 tok/s	MLX	ref
M4 Max (40-core GPU, 128 GB)	Llama 3.1 8B	Q4_K - Medium	—	51.6 tok/s	617.6 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Llama 3.1 8B	Q4_K - Medium	—	47.1 tok/s	557.1 tok/s	llamafile	ref
M4 Max (40-core GPU, 48 GB)	Qwen 2.5 14B	Q4_K - Medium	—	30.1 tok/s	347.0 tok/s	llamafile	ref
M4 Max (40-core GPU, 128 GB)	Qwen 2.5 14B	Q4_K - Medium	—	28.7 tok/s	326.6 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Qwen 2.5 14B	Q4_K - Medium	—	25.9 tok/s	286.7 tok/s	llamafile	ref
M4 Max (40-core GPU, 64 GB)	Qwen 3 32B	Q4_K - Medium	20.0 GB	22.0 tok/s	—	Ollama	Lab

Models tested on M4 Max (40-core GPU)

Llama 3.1 8B Llama 3.2 1B Qwen 3 4B Qwen 3 30B-A3B Qwen 3 32B Qwen 2.5 14B

Data

benchmarks.json — full dataset · chips.json — chip summaries · benchmarks.csv — CSV export

Data from in-house lab measurements plus community-published benchmarks. See all chip families →