readygpu

Mac Studio M4 Max 64GB vs NVIDIA RTX A6000

For running models locally, priced in the UK. At 128k context the first runs 60 models fully in memory, the second 60.

In short

The Mac Studio M4 Max 64GB has more memory for models: 64 GB against 48 GB. At 128k context both run the same 60 models we track fully in memory. On the models both run fully in memory, the NVIDIA RTX A6000 is about 1.3× as fast as the Mac Studio M4 Max 64GB (estimated, median across models). In the UK the Mac Studio M4 Max 64GB costs £3,399 at eBay and the NVIDIA RTX A6000 £3,999 at eBay (used): the Mac Studio M4 Max 64GB is £600 cheaper.

Updated . Speeds are estimates; how they are worked out.

Mac Studio M4 Max 64GB

£3,399

eBay

Add to build
NVIDIA RTX A6000

£3,999

eBay, used (asking)

Add to build

Specifications

SpecMac Studio M4 Max 64GBNVIDIA RTX A6000
ChipApple M4 Max (40-core GPU)–
Memory bus512-bit384-bit
Memory bandwidth546 GB/s768 GB/s
FP16 compute16.16 TFLOPS38.71 TFLOPS
SoftwareMetalCUDA
Unified memory64 GB–
GPU can use by default48 GB–
GPU can use, tuned56 GB–
Operating systemmacos–
ProcessorM4 Max 16-core–
Storage1000 GB–
GPU–RTX A6000
Memory–48 GB GDDR6
Board power–300 W
PCIe generation–Gen 4
PCIe lanes–x16

What each runs

Estimated decode speed with the most precise quant that fits each one est.

ModelMac Studio M4 Max 64GBNVIDIA RTX A6000
Qwen 3.8 27B9tok/sQ8_013tok/sQ8_0
Qwen 3.5 122B-A10B32tok/sUD-Q2_K_XL43tok/sUD-Q2_K_XL
Qwen 3.6 27B10tok/sQ8_013tok/sQ8_0
Gemma 4 31B8tok/sQ8_011tok/sQ8_0
Mistral Small 4 119B40tok/sUD-Q2_K_XL53tok/sUD-Q2_K_XL
Qwen 3.6 35B-A3B37tok/sQ8_049tok/sQ8_0
Qwen 3.5 27B10tok/sQ8_013tok/sQ8_0
Gemma 4 26B-A4B30tok/sQ8_041tok/sQ8_0
Qwen 3.5 35B-A3B37tok/sQ8_049tok/sQ8_0
GLM 4.7 Flash17tok/sQ8_023tok/sQ8_0
Devstral Small 2 24B8tok/sQ8_010tok/sQ8_0
Nemotron 3.5 Lightning 30B-A3B51tok/sUD-Q6_K_XL68tok/sUD-Q6_K_XL
Kimi Linear 48B-A3B60tok/sQ6_K80tok/sQ6_K
gpt-oss 20B38tok/sMXFP451tok/sMXFP4
Qwen3 Coder 30B-A3B13tok/sQ8_018tok/sQ8_0
Gemma 4 12B23tok/sQ8_031tok/sQ8_0
Qwen 3.5 9B26tok/sQ8_034tok/sQ8_0
Gemma 3 27B9tok/sQ8_012tok/sQ8_0
Mistral Small 3.2 24B8tok/sQ8_010tok/sQ8_0
Ministral 3 14B10tok/sQ8_013tok/sQ8_0
Qwen3 14B10tok/sQ8_013tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B13tok/sQ8_017tok/sQ8_0
Qwen3 8B13tok/sQ8_017tok/sQ8_0
Phi-4 14B8tok/sQ8_011tok/sQ8_0
Qwen 3.5 4B40tok/sQ8_054tok/sQ8_0
Ministral 3 8B13tok/sQ8_018tok/sQ8_0
Granite 4.2 8B12tok/sQ8_015tok/sQ8_0
Gemma 4 E4B30tok/sQ8_040tok/sQ8_0
LFM2.5 8B-A1B28tok/sQ8_038tok/sQ8_0
Llama 3.1 8B15tok/sQ6_K20tok/sQ6_K
Qwen2.5 0.5B166tok/sQ8_0222tok/sQ8_0
Qwen3.5 0.8B145tok/sQ8_0195tok/sQ8_0
Qwen3.5 2B116tok/sQ5_K_M155tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B63tok/sQ8_084tok/sQ8_0
Mellum2.1 12B A2.5B Thinking47tok/sQ8_062tok/sQ8_0
Spark X2.5 4B45tok/smlx-4bit60tok/smlx-4bit
MiniCPM5 2B43tok/sQ8_057tok/sQ8_0
FrogNano 4B 260940tok/sQ8_053tok/sQ8_0
LFM2.5 2.6B33tok/sQ8_044tok/sQ8_0
LFM2.5 VL 3B33tok/sQ8_044tok/sQ8_0
Kimi VL A3B Thinking 250630tok/sQ8_040tok/sQ8_0
Kimi VL A3B29tok/smlx-8bit39tok/smlx-8bit
GLM 4 9B 041426tok/smlx-5bit35tok/smlx-5bit
granite vision 4.1 4b25tok/sQ8_033tok/sQ8_0
granite 4.2 3b24tok/sQ8_033tok/sQ8_0
GLM 4.6V Flash23tok/sQ8_031tok/sQ8_0
DeepSeek R1 Distill Qwen 7B23tok/sQ8_031tok/sQ8_0
GLM 4.1V 9B Thinking21tok/smlx-8bit28tok/smlx-8bit
Ministral 3 3B Instruct 251220tok/sQ8_027tok/sQ8_0
LFM2.5 24B A2B17tok/sQ8_022tok/sQ8_0
DeepSeek R1 Distill Llama 8B14tok/sQ8_019tok/sQ8_0
Ministral 3 8B Reasoning 251213tok/sQ8_018tok/sQ8_0
Qwen3.8 27B OBLITERATED12tok/smlx-4bit16tok/smlx-4bit
Muse Glimmer 30B10tok/sQ8_014tok/sQ8_0
Ministral 3 14B Reasoning 251210tok/sQ8_013tok/sQ8_0
DeepSeek R1 Distill Qwen 14B9tok/sQ8_012tok/sQ8_0
Devstral Small 25078tok/sQ8_010tok/sQ8_0
Magistral Small 25068tok/sQ8_010tok/sQ8_0
Magistral Small 25078tok/sQ8_010tok/sQ8_0
Magistral Small 25098tok/sQ8_010tok/sQ8_0
Nemotron 3 Super 120B-A12Bwon't runoffload
Qwen3 Coder Nextwon't runoffload
gpt-oss 120Bwon't runoffload
GLM 4.5 Airwon't runoffload
Qwen3 Next 80B-A3Bwon't runoffload
Qwen3 32Bwon't runoffload
DeepSeek R1 Distill Qwen 32Bwon't runoffload
granite 4.2 30bwon't runoffload
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
Qwen 3.8 Flash Nextwon't runwon't run
MiniMax M2.7won't runwon't run
Llama 3.3 70Bwon't runwon't run
DeepSeek R1won't runwon't run
DeepSeek R1 0528won't runwon't run
DeepSeek R1 Distill Llama 70Bwon't runwon't run
DeepSeek R1 Zerowon't runwon't run
DeepSeek V3won't runwon't run
DeepSeek V3 0324won't runwon't run
DeepSeek V3.1won't runwon't run
DeepSeek V3.1 Terminuswon't runwon't run
DeepSeek V3.2won't runwon't run
DeepSeek V4 Flash 0731won't runwon't run
DeepSeek V4 Flash Vision Expwon't runwon't run
Devstral 2 123B Instruct 2512won't runwon't run
GLM 4.5won't runwon't run
GLM 4.5Vwon't runwon't run
GLM 4.6won't runwon't run
GLM 4.6Vwon't runwon't run
GLM 4.7won't runwon't run
GLM 5won't runwon't run
GLM 5.1won't runwon't run
GLM 5.2won't runwon't run
GLM 5.3won't runwon't run
Kimi Dev 72Bwon't runwon't run
Kimi K2won't runwon't run
Kimi K2 Instruct 0905won't runwon't run
Kimi K2 Thinkingwon't runwon't run
Kimi K2.5won't runwon't run
MiniMax M2won't runwon't run
MiniMax M2.1won't runwon't run
MiniMax M2.5won't runwon't run
MiniMax M3won't runwon't run
Mistral Medium 3.5 128Bwon't runwon't run

Questions

Which is better for local AI, the Mac Studio M4 Max 64GB or the NVIDIA RTX A6000?
Choose the NVIDIA RTX A6000 for speed. Choose the Mac Studio M4 Max 64GB to spend less.
Which can run bigger models, the Mac Studio M4 Max 64GB or the NVIDIA RTX A6000?
At 128k context both run the same 60 models we track fully in memory.
Which is faster for local AI, the Mac Studio M4 Max 64GB or the NVIDIA RTX A6000?
On the models both run fully in memory, the NVIDIA RTX A6000 is about 1.3× as fast as the Mac Studio M4 Max 64GB (estimated, median across models).
What is the price difference between the Mac Studio M4 Max 64GB and the NVIDIA RTX A6000?
In the UK the Mac Studio M4 Max 64GB costs £3,399 at eBay and the NVIDIA RTX A6000 £3,999 at eBay (used): the Mac Studio M4 Max 64GB is £600 cheaper.

More comparisons

More hardware for local AI