DeepSeek on Apple M4

Can I run DeepSeek-R1 671B-A37B on an Mac Mini M4 Pro 48GB?

Not at Q4_K_M — it needs 380.1 GB against 36.0 GB available. No quantisation of this model fits on one card.

Won't fit

380.1 GB of 36.0 GB · 1056%
0395 GB
Weights 379.0 GB
KV cache 0.5 GB
Runtime overhead 0.6 GB
Over the limit 344.1 GB

Short by 344.1 GB. You can run it with 5 of 61 layers on the Mac Mini M4 Pro 48GB and the rest in system RAM, at roughly 2.04 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.

Generation2.04tok/s
Prompt processing96.5tok/s
Max context0tokens
KV per 1K tokens0GB

Every quantisation of DeepSeek-R1 671B-A37B 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 1249.8 GB 1250.9 GB Won't fit 0.60
INT8 / W8A8 8.50 663.6 GB 664.7 GB Won't fit 1.15
Q8_0 (GGUF) 8.50 663.6 GB 664.7 GB Won't fit 1.15
FP8 (E4M3) 8.00 624.9 GB 626.0 GB Won't fit 1.22
Q6_K 6.56 512.4 GB 513.5 GB Won't fit 1.50
Q5_K_M 5.67 442.9 GB 444.0 GB Won't fit 1.73
Q5_K_S 5.52 431.2 GB 432.3 GB Won't fit 1.78
Q4_K_M 4.85 379.0 GB 380.1 GB Won't fit 2.04
Q4_K_S 4.58 358.0 GB 359.1 GB Won't fit 2.16
Q4_0 4.55 355.6 GB 356.7 GB Won't fit 2.17
AWQ 4-bit 4.25 334.5 GB 335.6 GB Won't fit 2.34
GPTQ 4-bit 4.25 334.5 GB 335.6 GB Won't fit 2.34
MXFP4 4.25 334.5 GB 335.6 GB Won't fit 2.34
IQ4_XS 4.25 332.3 GB 333.4 GB Won't fit 2.35
Q3_K_M 3.91 305.8 GB 306.9 GB Won't fit 2.55
IQ3_M 3.70 289.4 GB 290.5 GB Won't fit 2.72
IQ3_XXS 3.06 239.6 GB 240.7 GB Won't fit 3.31
Q2_K 2.63 206.1 GB 207.2 GB Won't fit 3.93
IQ2_XXS 2.06 161.7 GB 162.8 GB Won't fit 5.15
IQ1_M 1.75 137.5 GB 138.6 GB Won't fit 6.17

Also worth checking