MacBook Pro M3 Max 48GB

Apple · M3 Max · 48GB Unified Memory · Can run 42 models

Buy Apple
Manufacturer Apple
Unified Mem 48 GB
Chip M3 Max
CPU Cores 16
GPU Cores 40
Neural Engine 16
Bandwidth 400 GB/s
MSRP $3,499
Released Nov 7, 2023

AI Notes

The MacBook Pro M3 Max 48GB is a powerhouse for local AI inference. With 400 GB/s bandwidth and 48GB unified memory, it runs 30B models at fast speeds and can handle 70B models with aggressive quantization. The massive bandwidth advantage over Pro chips translates to dramatically faster token generation.

Compatible Models

Model Parameters Best Quant VRAM Used Fit Est. Speed
Gemma 3 1B 1B Q8_0 2 GB Runs ~200 tok/s
Llama 3.2 1B 1B Q8_0 3 GB Runs ~133 tok/s
DeepSeek R1 1.5B 1.5B Q8_0 3 GB Runs ~133 tok/s
Gemma 2 2B 2B Q8_0 4 GB Runs ~100 tok/s
Llama 3.2 3B 3B Q8_0 5 GB Runs ~80 tok/s
Phi-3 Mini 3.8B 3.8B Q8_0 5.8 GB Runs ~69 tok/s
Phi-4 Mini 3.8B 3.8B Q4_K_M 4.5 GB Runs ~89 tok/s
Gemma 3 4B 4B Q4_K_M 5 GB Runs ~80 tok/s
Qwen 3 4B 4B Q4_K_M 4.5 GB Runs ~89 tok/s
DeepSeek R1 7B 7B Q8_0 9 GB Runs ~44 tok/s
Mistral 7B 7B Q8_0 9 GB Runs ~44 tok/s
Qwen 2.5 7B 7B Q8_0 9 GB Runs ~44 tok/s
Qwen 2.5 Coder 7B 7B Q8_0 9 GB Runs ~44 tok/s
DeepSeek R1 8B 8B Q4_K_M 7.5 GB Runs ~53 tok/s
Llama 3.1 8B 8B Q8_0 10 GB Runs ~40 tok/s
Qwen 3 8B 8B Q4_K_M 7.5 GB Runs ~53 tok/s
Gemma 2 9B 9B Q8_0 11 GB Runs ~36 tok/s
Gemma 3 12B 12B Q4_K_M 10.5 GB Runs ~38 tok/s
Mistral Nemo 12B 12B Q4_K_M 9.5 GB Runs ~42 tok/s
DeepSeek R1 14B 14B Q4_K_M 9.9 GB Runs ~40 tok/s
Phi-4 14B 14B Q4_K_M 9.9 GB Runs ~40 tok/s
Qwen 2.5 14B 14B Q4_K_M 9.9 GB Runs ~40 tok/s
Qwen 2.5 Coder 14B 14B Q4_K_M 12 GB Runs ~33 tok/s
Qwen 3 14B 14B Q4_K_M 12 GB Runs ~33 tok/s
StarCoder2 15B 15B Q8_0 17 GB Runs ~24 tok/s
Codestral 22B 22B Q4_K_M 14.7 GB Runs ~27 tok/s
Devstral 24B 24B Q4_K_M 17 GB Runs ~24 tok/s
Mistral Small 3.1 24B 24B Q4_K_M 18 GB Runs ~22 tok/s
Gemma 2 27B 27B Q4_K_M 17.7 GB Runs ~23 tok/s
Gemma 3 27B 27B Q4_K_M 20 GB Runs ~20 tok/s
Qwen 3 30B-A3B (MoE) 30B Q4_K_M 22 GB Runs ~18 tok/s
DeepSeek R1 32B 32B Q4_K_M 20.7 GB Runs ~19 tok/s
Qwen 2.5 32B 32B Q4_K_M 20.7 GB Runs ~19 tok/s
Qwen 2.5 Coder 32B 32B Q4_K_M 23 GB Runs ~17 tok/s
Qwen 3 32B 32B Q4_K_M 23 GB Runs ~17 tok/s
Command R 35B 35B Q4_K_M 22.5 GB Runs ~18 tok/s
Mixtral 8x7B 47B Q4_K_M 29.7 GB Runs ~13 tok/s
DeepSeek R1 70B 70B Q4_K_M 43.5 GB Runs (tight) ~9 tok/s
Llama 3.1 70B 70B Q4_K_M 43.5 GB Runs (tight) ~9 tok/s
Llama 3.3 70B 70B Q4_K_M 43.5 GB Runs (tight) ~9 tok/s
Qwen 2.5 72B 72B Q4_K_M 44.7 GB Runs (tight) ~9 tok/s
Llama 4 Scout (109B/17B active) 109B Q4_K_M 72 GB CPU Offload ~6 tok/s
1 model(s) are too large for this hardware.