Apple · M4 · unified memory

What runs on a Mac Mini M4 24GB

24 GB at 120 GB/s. The largest model that fits at Q4_K_M is Gemma 3 27B, at roughly 4.72 tokens per second. Unified memory means the operating system takes a share, so budget around three quarters of the total.

Runs comfortably

6.1 GB of 18.0 GB · 34%
018 GB
Weights 4.6 GB
KV cache 1.0 GB
Runtime overhead 0.5 GB

Llama 3.1 8B at Q4_K_M leaves 11.9 GB spare on a Mac Mini M4 24GB. There is room to raise the context length or move up a quantisation level.

Generation15.1tok/s
Prompt processing235tok/s
Max context103Ktokens
KV per 1K tokens0GB

Every quantisation of Llama 3.1 8B on a Mac Mini M4 24GB

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 15.0 GB 16.5 GB Fits, but tight 20K 4.99
INT8 / W8A8 8.50 7.7 GB 9.2 GB Runs comfortably 78K 9.39
Q8_0 (GGUF) 8.50 7.7 GB 9.2 GB Runs comfortably 78K 9.39
FP8 (E4M3) 8.00 7.5 GB 9.0 GB Runs comfortably 80K 9.66
Q6_K 6.56 6.1 GB 7.7 GB Runs comfortably 91K 11.6
AWQ 4-bit 4.25 5.4 GB 6.9 GB Runs comfortably 97K 13.0
GPTQ 4-bit 4.25 5.4 GB 6.9 GB Runs comfortably 97K 13.0
MXFP4 4.25 5.4 GB 6.9 GB Runs comfortably 97K 13.0
Q5_K_M 5.67 5.3 GB 6.8 GB Runs comfortably 97K 13.3
Q5_K_S 5.52 5.2 GB 6.7 GB Runs comfortably 99K 13.6
Q4_K_M 4.85 4.6 GB 6.1 GB Runs comfortably 103K 15.1
Q4_K_S 4.58 4.4 GB 5.9 GB Runs comfortably 105K 15.7
Q4_0 4.55 4.4 GB 5.9 GB Runs comfortably 105K 15.8
IQ4_XS 4.25 4.1 GB 5.6 GB Runs comfortably 107K 16.6
Q3_K_M 3.91 3.8 GB 5.4 GB Runs comfortably 109K 17.7
IQ3_M 3.70 3.7 GB 5.2 GB Runs comfortably 110K 18.4
IQ3_XXS 3.06 3.2 GB 4.7 GB Runs comfortably 115K 21.0
Q2_K 2.63 2.8 GB 4.3 GB Runs comfortably 117K 23.2
IQ2_XXS 2.06 2.3 GB 3.9 GB Runs comfortably 121K 27.0
IQ1_M 1.75 2.1 GB 3.6 GB Runs comfortably 123K 29.6

Models on a Mac Mini M4 24GB

Q4_K_M at 8K context. Bandwidth sets the speed; capacity sets the ceiling.

ModelParamsWeightsVerdictMax ctxtok/s
DeepSeek-R1 671B-A37B 671B 379.0 GB Won't fit 1.95
Qwen3 235B-A22B 235.1B 132.8 GB Won't fit 3.16
gpt-oss 120B-A5.1B 116.8B 66.0 GB Won't fit 14.0
Llama 4 Scout 109B-A17B 109B 61.7 GB Won't fit 4.37
Qwen2.5 72B 72.7B 41.2 GB Won't fit 1.16
Llama 3.1 70B 70.5B 40.0 GB Won't fit 1.20
Mixtral 8x7B 46.7B 26.4 GB Won't fit 6.79
Command R 35B 35.0B 20.1 GB Won't fit 1.96
Yi-1.5 34B 34.4B 19.5 GB Won't fit 3.17
Qwen3 32B 32.8B 18.6 GB Won't fit 3.41
Qwen2.5-Coder 32B 32.8B 18.6 GB Won't fit 3.41
DeepSeek-R1-Distill-Qwen 32B 32.8B 18.6 GB Won't fit 3.41
Qwen3 30B-A3B 30.5B 17.3 GB Won't fit 2K 25.6
Gemma 3 27B 27.4B 15.6 GB Fits, but tight 17K 4.72
Mistral Small 24B 23.6B 13.4 GB Runs comfortably 26K 5.49
gpt-oss 20B-A3.6B 20.9B 11.9 GB Runs comfortably 120K 28.8
Qwen3 14B 14.8B 8.5 GB Runs comfortably 58K 8.48
Phi-4 14B 14.7B 8.4 GB Runs comfortably 16K 8.41
Mistral NeMo 12B 12.3B 7.0 GB Runs comfortably 67K 10.1
Gemma 3 12B 12.2B 7.0 GB Runs comfortably 128K 10.2
GLM-4 9B 9.4B 5.4 GB Runs comfortably 128K 13.8
Qwen3 8B 8.2B 4.7 GB Runs comfortably 91K 14.6
Llama 3.1 8B 8.0B 4.6 GB Runs comfortably 103K 15.1
DeepSeek-R1-Distill-Llama 8B 8.0B 4.6 GB Runs comfortably 103K 15.1
Qwen2.5 7B 7.6B 4.4 GB Runs comfortably 128K 16.7
Mistral 7B v0.3 7.3B 4.1 GB Runs comfortably 32K 16.7
Gemma 3 4B 4.3B 2.5 GB Runs comfortably 128K 28.6
Qwen3 4B 4.0B 2.3 GB Runs comfortably 108K 26.8
Llama 3.2 3B 3.2B 1.8 GB Runs comfortably 128K 33.7
Llama 3.2 1B 1.2B 0.7 GB Runs comfortably 128K 90.6

Direct answers