M1 Max (24-core GPU) — LLM Benchmarks

Measured LLM inference benchmarks for M1 Max (24-core GPU) across all RAM configurations (32 GB, 64 GB). 5 benchmark rows across 3 models. Compare how RAM affects throughput. Real runs, not estimates.

5Benchmark rows

3Models tested

2RAM configurations

105.7Fastest 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.

32 GBM1 Max (24-core GPU, 32 GB)2 rows · 2 models93.9 tok/s peak 64 GBM1 Max (24-core GPU, 64 GB)3 rows · 3 models105.7 tok/s peak

All benchmark rows — M1 Max (24-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
M1 Max (24-core GPU, 64 GB)	Llama 3.2 1B Instruct	Q4_K - Medium	—	105.7 tok/s	1669.5 tok/s	—	ref
M1 Max (24-core GPU, 32 GB)	Llama 3.2 1B Instruct	Q4_K - Medium	—	93.9 tok/s	1582.0 tok/s	—	ref
M1 Max (24-core GPU, 64 GB)	Llama 3.1 8B Instruct	Q4_K - Medium	—	32.1 tok/s	306.2 tok/s	—	ref
M1 Max (24-core GPU, 32 GB)	Qwen 2.5 14B Instruct	Q4_K - Medium	—	17.4 tok/s	155.9 tok/s	—	ref
M1 Max (24-core GPU, 64 GB)	Qwen 2.5 14B Instruct	Q4_K - Medium	—	15.1 tok/s	140.2 tok/s	—	ref

Models tested on M1 Max (24-core GPU)

Llama 3.1 8B Instruct Llama 3.2 1B Instruct Qwen 2.5 14B Instruct

Data

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

Data sourced from factory lab measurements and community reference runs. See all chips →