Apple · M3 · unified memory

What runs on a Mac Studio M3 Ultra 256GB

256 GB at 819 GB/s. The largest model that fits at Q4_K_M is Qwen3 235B-A22B, at roughly 24.7 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 192.0 GB · 3%
0192 GB
Weights 4.6 GB
KV cache 1.0 GB
Runtime overhead 0.5 GB

Llama 3.1 8B at Q4_K_M leaves 185.9 GB spare on a Mac Studio M3 Ultra 256GB. There is room to raise the context length or move up a quantisation level.

Generation90.9tok/s
Prompt processing1491tok/s
Max context128Ktokens
KV per 1K tokens0GB

Every quantisation of Llama 3.1 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 15.0 GB 16.5 GB Runs comfortably 128K 30.4
INT8 / W8A8 8.50 7.7 GB 9.2 GB Runs comfortably 128K 57.0
Q8_0 (GGUF) 8.50 7.7 GB 9.2 GB Runs comfortably 128K 57.0
FP8 (E4M3) 8.00 7.5 GB 9.0 GB Runs comfortably 128K 58.6
Q6_K 6.56 6.1 GB 7.7 GB Runs comfortably 128K 70.4
AWQ 4-bit 4.25 5.4 GB 6.9 GB Runs comfortably 128K 78.9
GPTQ 4-bit 4.25 5.4 GB 6.9 GB Runs comfortably 128K 78.9
MXFP4 4.25 5.4 GB 6.9 GB Runs comfortably 128K 78.9
Q5_K_M 5.67 5.3 GB 6.8 GB Runs comfortably 128K 80.3
Q5_K_S 5.52 5.2 GB 6.7 GB Runs comfortably 128K 82.3
Q4_K_M 4.85 4.6 GB 6.1 GB Runs comfortably 128K 90.9
Q4_K_S 4.58 4.4 GB 5.9 GB Runs comfortably 128K 94.9
Q4_0 4.55 4.4 GB 5.9 GB Runs comfortably 128K 95.4
IQ4_XS 4.25 4.1 GB 5.6 GB Runs comfortably 128K 100
Q3_K_M 3.91 3.8 GB 5.4 GB Runs comfortably 128K 107
IQ3_M 3.70 3.7 GB 5.2 GB Runs comfortably 128K 111
IQ3_XXS 3.06 3.2 GB 4.7 GB Runs comfortably 128K 126
Q2_K 2.63 2.8 GB 4.3 GB Runs comfortably 128K 139
IQ2_XXS 2.06 2.3 GB 3.9 GB Runs comfortably 128K 161
IQ1_M 1.75 2.1 GB 3.6 GB Runs comfortably 128K 176

Models on a Mac Studio M3 Ultra 256GB

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 — 3.10
Qwen3 235B-A22B 235.1B 132.8 GB Runs comfortably 128K 24.7
gpt-oss 120B-A5.1B 116.8B 66.0 GB Runs comfortably 128K 107
Llama 4 Scout 109B-A17B 109B 61.7 GB Runs comfortably 692K 38.5
Qwen2.5 72B 72.7B 41.2 GB Runs comfortably 128K 11.1
Llama 3.1 70B 70.5B 40.0 GB Runs comfortably 128K 11.4
Mixtral 8x7B 46.7B 26.4 GB Runs comfortably 32K 50.5
Command R 35B 35.0B 20.1 GB Runs comfortably 128K 18.8
Yi-1.5 34B 34.4B 19.5 GB Runs comfortably 32K 23.0
Qwen3 32B 32.8B 18.6 GB Runs comfortably 128K 23.9
Qwen2.5-Coder 32B 32.8B 18.6 GB Runs comfortably 128K 23.9
DeepSeek-R1-Distill-Qwen 32B 32.8B 18.6 GB Runs comfortably 128K 23.9
Qwen3 30B-A3B 30.5B 17.3 GB Runs comfortably 128K 104
Gemma 3 27B 27.4B 15.6 GB Runs comfortably 128K 28.7
Mistral Small 24B 23.6B 13.4 GB Runs comfortably 32K 33.4
gpt-oss 20B-A3.6B 20.9B 11.9 GB Runs comfortably 128K 154
Qwen3 14B 14.8B 8.5 GB Runs comfortably 128K 51.5
Phi-4 14B 14.7B 8.4 GB Runs comfortably 16K 51.0
Mistral NeMo 12B 12.3B 7.0 GB Runs comfortably 128K 61.0
Gemma 3 12B 12.2B 7.0 GB Runs comfortably 128K 61.8
GLM-4 9B 9.4B 5.4 GB Runs comfortably 128K 83.4
Qwen3 8B 8.2B 4.7 GB Runs comfortably 128K 87.9
Llama 3.1 8B 8.0B 4.6 GB Runs comfortably 128K 90.9
DeepSeek-R1-Distill-Llama 8B 8.0B 4.6 GB Runs comfortably 128K 90.9
Qwen2.5 7B 7.6B 4.4 GB Runs comfortably 128K 101
Mistral 7B v0.3 7.3B 4.1 GB Runs comfortably 32K 101
Gemma 3 4B 4.3B 2.5 GB Runs comfortably 128K 170
Qwen3 4B 4.0B 2.3 GB Runs comfortably 128K 160
Llama 3.2 3B 3.2B 1.8 GB Runs comfortably 128K 201
Llama 3.2 1B 1.2B 0.7 GB Runs comfortably 128K 531

Direct answers

Llama 3.1 8B on Mac Studio M3 Ultra 256GB
8.0BLlama 3.1 70B on Mac Studio M3 Ultra 256GB
70.5BLlama 3.2 3B on Mac Studio M3 Ultra 256GB
3.2BLlama 3.2 1B on Mac Studio M3 Ultra 256GB
1.2BLlama 4 Scout 109B-A17B on Mac Studio M3 Ultra 256GB
109BQwen3 8B on Mac Studio M3 Ultra 256GB
8.2BQwen3 14B on Mac Studio M3 Ultra 256GB
14.8BQwen3 32B on Mac Studio M3 Ultra 256GB
32.8B