RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run an LLM on my Mac?

Apple Silicon shares memory between the CPU and GPU. By default macOS lets the GPU use about 67% of RAM on machines under 64 GB, and about 75% on 64 GB and larger — less than the 95% we assume for a discrete GPU with the same advertised memory. The tables below use those default limits.

Macs we track

MacUnified memoryUsable (default)Models that fit (8K)
Mac M1 16GB16 GB10.7 GB165
Mac M4 16GB16 GB10.7 GB165
Mac M2 24GB24 GB16.1 GB199
Mac M3 24GB24 GB16.1 GB199
Mac M4 Pro 24GB24 GB16.1 GB199
Mac M1 Pro 32GB32 GB21.4 GB202
Mac M4 32GB32 GB21.4 GB202
Mac M3 Pro 36GB36 GB24.1 GB202
Mac M4 Pro 48GB48 GB32.2 GB207
Mac M1 Max 64GB64 GB48.0 GB215
Mac M4 Max 64GB64 GB48.0 GB215
Mac M4 Pro 64GB64 GB48.0 GB215
Mac M2 Max 96GB96 GB72.0 GB219
Mac M3 Max 128GB128 GB96.0 GB227
Mac M4 Max 128GB128 GB96.0 GB227
Mac M2 Ultra 192GB192 GB144.0 GB245
Mac M3 Ultra 256GB256 GB192.0 GB248
Mac M3 Ultra 512GB512 GB384.0 GB278

How many models fit at each memory size

Unified memoryUsable (default)Models that fitLargest that fitsExample machines
16 GB10.7 GB165Trinity MiniMac M1 16GB, Mac M4 16GB
24 GB16.1 GB199flintMac M2 24GB, Mac M3 24GB, Mac M4 Pro 24GB
32 GB21.4 GB202Kimi Linear 48B A3B InstructMac M1 Pro 32GB, Mac M4 32GB
36 GB24.1 GB202Kimi Linear 48B A3B InstructMac M3 Pro 36GB
48 GB32.2 GB207Hunyuan A13B InstructMac M4 Pro 48GB
64 GB48.0 GB215Ling 3.0 flash SanteMac M1 Max 64GB, Mac M4 Max 64GB, Mac M4 Pro 64GB
96 GB72.0 GB219Step 3.5 FlashMac M2 Max 96GB
128 GB96.0 GB227Intern S1Mac M3 Max 128GB, Mac M4 Max 128GB
192 GB144.0 GB245Llama 4 Maverick 17B 128E InstructMac M2 Ultra 192GB
256 GB192.0 GB248MiniMax M1 40kMac M3 Ultra 256GB
512 GB384.0 GB278Kimi K2.5Mac M3 Ultra 512GB

Popular models that fit a 16 GB Mac

Best quality quant that fits the Mac M4 16GB at 8K context, sorted by Hugging Face downloads. Full list for this Mac.

ModelParamsBest quantNeedsMax tok/s (est.)
Qwen3 0.6B752MF162.7 GB100
Qwen3 14B15BQ3_K_M8.7 GB17
Qwen3 VL 8B Instruct8.8BQ6_K8.5 GB17
Qwen2.5 7B Instruct7.6BQ8_08.7 GB15
Qwen3.5 9B9.7BQ6_K8.1 GB16
Qwen3 8B8.2BQ6_K8.1 GB18
Unlimited OCR3.3BF166.6 GB60
Qwen2.5 VL 7B Instruct8.3BQ8_08.7 GB15
gemma 4 26B A4B it26BQ2_K10.4 GB80
Qwen2.5 1.5B Instruct1.5BF163.8 GB39
Qwen3.5 2B2.3BF164.5 GB32
Qwen3.5 4B4.7BF169.0 GB14
Llama 3.1 8B Instruct8.0BQ8_09.6 GB14
Qwen2.5 0.5B Instruct494MF161.7 GB121
Qwen3 VL 4B Instruct4.4BF169.3 GB15
Qwen3 4B4.0BF169.3 GB15
Qwen3 1.7B2.0BF164.8 GB35
Qwen3.5 0.8B873MF162.4 GB79
Qwen3 4B Instruct 25074.0BF169.3 GB15
Qwen2.5 3B Instruct3.1BF166.7 GB19

Questions

Can I run an LLM on a Mac?

Yes. We track 285 open models across 18 Apple Silicon Macs. Open your exact machine above for the full list, best quant and an estimated speed ceiling.

Why does a 16 GB Mac fit fewer models than a 16 GB NVIDIA GPU?

macOS lets the GPU wire about 67% of unified memory on machines under 64 GB, and about 75% on 64 GB and larger, by default. A discrete GPU is modelled at 95% of VRAM. You can raise the Mac limit with the iogpu.wired_limit_mb setting; the numbers here stay on the defaults so they match a stock machine.

Looking for a discrete GPU instead? See models by VRAM size or the full hardware list. Methodology · Report an error.