Qwen on Apple M4

Can I run Qwen3 32B on an Mac Mini M4 Pro 48GB?

Yes. Qwen3 32B at Q4_K_M uses 21.2 GB of the 36.0 GB available on a Mac Mini M4 Pro 48GB, and runs at about 8.93 tokens per second. You can push the context to 67K.

Runs comfortably

21.2 GB of 36.0 GB · 59%
036 GB
Weights 18.6 GB
KV cache 2.0 GB
Runtime overhead 0.5 GB

Qwen3 32B at Q4_K_M leaves 14.8 GB spare on a Mac Mini M4 Pro 48GB. There is room to raise the context length or move up a quantisation level.

Generation8.93tok/s
Prompt processing109tok/s
Max context67Ktokens
KV per 1K tokens0GB

Every quantisation of Qwen3 32B on a Mac Mini M4 Pro 48GB

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

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 61.0 GB 63.6 GB Won't fit — 1.14
INT8 / W8A8 8.50 32.1 GB 34.6 GB Fits, but tight 14K 5.30
Q8_0 (GGUF) 8.50 32.1 GB 34.6 GB Fits, but tight 14K 5.30
FP8 (E4M3) 8.00 30.5 GB 33.0 GB Fits, but tight 20K 5.56
Q6_K 6.56 25.0 GB 27.6 GB Runs comfortably 42K 6.73
Q5_K_M 5.67 21.6 GB 24.2 GB Runs comfortably 55K 7.74
Q5_K_S 5.52 21.1 GB 23.6 GB Runs comfortably 58K 7.94
Q4_K_M 4.85 18.6 GB 21.2 GB Runs comfortably 67K 8.93
AWQ 4-bit 4.25 18.3 GB 20.9 GB Runs comfortably 68K 9.05
GPTQ 4-bit 4.25 18.3 GB 20.9 GB Runs comfortably 68K 9.05
MXFP4 4.25 18.3 GB 20.9 GB Runs comfortably 68K 9.05
Q4_K_S 4.58 17.6 GB 20.2 GB Runs comfortably 71K 9.39
Q4_0 4.55 17.5 GB 20.1 GB Runs comfortably 72K 9.45
IQ4_XS 4.25 16.4 GB 19.0 GB Runs comfortably 76K 10.0
Q3_K_M 3.91 15.2 GB 17.7 GB Runs comfortably 81K 10.8
IQ3_M 3.70 14.4 GB 17.0 GB Runs comfortably 84K 11.3
IQ3_XXS 3.06 12.1 GB 14.7 GB Runs comfortably 93K 13.3
Q2_K 2.63 10.6 GB 13.1 GB Runs comfortably 100K 15.1
IQ2_XXS 2.06 8.5 GB 11.0 GB Runs comfortably 108K 18.4
IQ1_M 1.75 7.4 GB 9.9 GB Runs comfortably 112K 20.9

Also worth checking

Qwen3 32B on every GPU
Full hardware tableEverything that runs on a Mac Mini M4 Pro 48GB
Full model tableWhy the KV cache is the part that bites
Guide