Qwen on Apple M3

Can I run Qwen3 8B on an Mac Studio M3 Ultra 256GB?

Yes. Qwen3 8B at Q4_K_M uses 6.4 GB of the 192.0 GB available on a Mac Studio M3 Ultra 256GB, and runs at about 87.9 tokens per second. You can push the context to 128K.

Runs comfortably

6.4 GB of 192.0 GB · 3%
0192 GB
Weights 4.7 GB
KV cache 1.1 GB
Runtime overhead 0.5 GB

Qwen3 8B at Q4_K_M leaves 185.6 GB spare on a Mac Studio M3 Ultra 256GB. There is room to raise the context length or move up a quantisation level.

Generation87.9tok/s
Prompt processing1462tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Qwen3 8B 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 15.3 GB 16.9 GB Runs comfortably 128K 29.7
INT8 / W8A8 8.50 7.8 GB 9.5 GB Runs comfortably 128K 55.8
Q8_0 (GGUF) 8.50 7.8 GB 9.5 GB Runs comfortably 128K 55.8
FP8 (E4M3) 8.00 7.6 GB 9.3 GB Runs comfortably 128K 57.1
Q6_K 6.56 6.3 GB 7.9 GB Runs comfortably 128K 68.4
AWQ 4-bit 4.25 5.8 GB 7.4 GB Runs comfortably 128K 73.7
GPTQ 4-bit 4.25 5.8 GB 7.4 GB Runs comfortably 128K 73.7
MXFP4 4.25 5.8 GB 7.4 GB Runs comfortably 128K 73.7
Q5_K_M 5.67 5.4 GB 7.1 GB Runs comfortably 128K 77.9
Q5_K_S 5.52 5.3 GB 6.9 GB Runs comfortably 128K 79.8
Q4_K_M 4.85 4.7 GB 6.4 GB Runs comfortably 128K 87.9
Q4_K_S 4.58 4.5 GB 6.1 GB Runs comfortably 128K 91.6
Q4_0 4.55 4.5 GB 6.1 GB Runs comfortably 128K 92.1
IQ4_XS 4.25 4.2 GB 5.9 GB Runs comfortably 128K 96.6
Q3_K_M 3.91 4.0 GB 5.6 GB Runs comfortably 128K 102
IQ3_M 3.70 3.8 GB 5.4 GB Runs comfortably 128K 106
IQ3_XXS 3.06 3.3 GB 4.9 GB Runs comfortably 128K 120
Q2_K 2.63 2.9 GB 4.6 GB Runs comfortably 128K 132
IQ2_XXS 2.06 2.5 GB 4.1 GB Runs comfortably 128K 151
IQ1_M 1.75 2.2 GB 3.9 GB Runs comfortably 128K 165

Also worth checking

Qwen3 8B 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