Machines / Mac mini M6 (16GB)

What LLMs can the Mac mini M6 (16GB) run?

The Mac mini M6 (16GB) has 16 GB of memory (about 11 GB usable by a model) and 170 GB/s of bandwidth. The largest model it runs at a comfortable 10+ tokens/sec is Phi 4 14B.

16 GBmemory

170 GB/sbandwidth

$899from

Check the Mac mini M6 (16GB) price →

Fast (20+ tok/s)

Qwen3.5 0.8B 120+DeepSeek-R1 1.5B 120+Qwen 1.8B 120+lfm 2.5 8B (A1B) 111 Llama 3.2 1B 105 Qwen 3 1.7B 97 Granite 3.1 moe 1B 97 minicpm v4.6 1B 85 Gemma 2 2B 85 codeGemma 2B 85 Gemma 2B 80 smollm 2 1.7B 76 starCoder 2 3B 76 falcon 3 1B 76 Qwen2.5 3B 72 Qwen2.5 Coder 3B 72 Qwen3.5 2B 72 Qwen 3-VL 2B 72 Llama 3.2 3B 68 orca mini 3B 68 Granite 3.1 moe 3B 68 falcon 3 3B 68 Granite 4.1 3B 65 Granite 4.2 3B 62 Phi 3 3.8B 62 Qwen 4B 59 Qwen 3 4B 52 Qwen2.5-VL 3B 43 Gemma 3 4B 41 Qwen 3-VL 4B 41 translateGemma 4B 41 Qwen3.5 4B 40 Llama 2 7B 36 Code Llama 7B 36 DeepSeek Coder 6.7B 36 Llama 2 uncensored 7B 36 orca mini 7B 33 Mistral 7B 31 Qwen2 7B 31 starCoder 2 7B 31 OLMo 2 7B 30 falcon 3 7B 30 DeepSeek-R1 7B 29 Qwen2.5 7B 29 Llama 3 8B 29 Qwen2.5 Coder 7B 29 LLaVA 7B 29 Llama 3.1 8B 28 Qwen 7B 28 dolPhin 3 8B 28 Gemma 7B 27 codeGemma 7B 27 DeepSeek-R1 8B 26 Qwen 3 8B 26 Granite 4.2 8B 26 Granite 4.1 8B 26 Gemma 2 9B 25 minicpm v 8B 25 LLaVA Llama3 8B 25 ornith 9B 24 Gemma 3n E2B 24 Qwen2.5-VL 7B 23 minicpm v4.5 8B 22 Qwen 3-VL 8B 22 falcon 3 10B 22 ornith 1.5 9B 21 Qwen3.5 9B 21

Comfortable (10-20 tok/s)

Mistral nemo 12B 19 Gemma 4 E2B 19 Llama 2 13B 18 Code Llama 13B 18 Gemma 3n E4B 18 Gemma 4 12B 18 Llama 3.2 Vision 11B 17 Phi 3 14B 17 orca mini 13B 17 LLaVA 13B 17 Gemma 3 12B 17 translateGemma 12B 17 Qwen 14B 17 OLMo 2 13B 16 DeepSeek Coder v2 16B 15 DeepSeek-R1 14B 15 Qwen2.5 14B 15 Qwen2.5 Coder 14B 15 Phi 4 14B 15

Slow but usable (5-10 tok/s)

Qwen 3 14B 8.8 Gemma 4 E4B 8.5 starCoder 2 15B 8.2

Won't fit