Qwen on Apple M3

Can I run Qwen3 4B on an Mac Studio M3 Ultra 256GB?

Yes. Qwen3 4B at Q4_K_M uses 3.9 GB of the 192.0 GB available on a Mac Studio M3 Ultra 256GB, and runs at about 160 tokens per second. You can push the context to 128K.

Runs comfortably

3.9 GB of 192.0 GB · 2%
0192 GB
Weights 2.3 GB
KV cache 1.1 GB
Runtime overhead 0.5 GB

Qwen3 4B at Q4_K_M leaves 188.1 GB spare on a Mac Studio M3 Ultra 256GB. There is room to raise the context length or move up a quantisation level.

Generation160tok/s
Prompt processing2978tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Qwen3 4B on a Mac Studio M3 Ultra 256GB

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

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 7.5 GB 9.1 GB Runs comfortably 128K 58.0
INT8 / W8A8 8.50 3.9 GB 5.5 GB Runs comfortably 128K 104
Q8_0 (GGUF) 8.50 3.9 GB 5.5 GB Runs comfortably 128K 104
FP8 (E4M3) 8.00 3.7 GB 5.4 GB Runs comfortably 128K 107
Q6_K 6.56 3.1 GB 4.7 GB Runs comfortably 128K 127
Q5_K_M 5.67 2.7 GB 4.3 GB Runs comfortably 128K 143
Q5_K_S 5.52 2.6 GB 4.2 GB Runs comfortably 128K 146
AWQ 4-bit 4.25 2.5 GB 4.1 GB Runs comfortably 128K 149
GPTQ 4-bit 4.25 2.5 GB 4.1 GB Runs comfortably 128K 149
MXFP4 4.25 2.5 GB 4.1 GB Runs comfortably 128K 149
Q4_K_M 4.85 2.3 GB 3.9 GB Runs comfortably 128K 160
Q4_K_S 4.58 2.2 GB 3.8 GB Runs comfortably 128K 166
Q4_0 4.55 2.2 GB 3.8 GB Runs comfortably 128K 167
IQ4_XS 4.25 2.0 GB 3.7 GB Runs comfortably 128K 175
Q3_K_M 3.91 1.9 GB 3.5 GB Runs comfortably 128K 184
IQ3_M 3.70 1.8 GB 3.4 GB Runs comfortably 128K 191
IQ3_XXS 3.06 1.5 GB 3.2 GB Runs comfortably 128K 214
Q2_K 2.63 1.4 GB 3.0 GB Runs comfortably 128K 234
IQ2_XXS 2.06 1.1 GB 2.7 GB Runs comfortably 128K 265
IQ1_M 1.75 1.0 GB 2.6 GB Runs comfortably 128K 286

Also worth checking

Qwen3 4B on every GPU
Full hardware tableEverything that runs on a Mac Studio M3 Ultra 256GB
Full model tableWhy the KV cache is the part that bites
Guide