← All categories
Apple unified memory
Macs give the GPU most of one large memory pool, run MLX and llama.cpp through Metal, and are quiet. Bandwidth rises steeply from Pro to Max to Ultra chips.
| Machine | Memory for models | Bandwidth | Price | Largest at 4-bit | gpt-oss-20b |
|---|---|---|---|---|---|
| Apple Mac Studio (M5 Max, 32-core GPU)Apple unified memory | 24 GBassumed | 460 GB/s | ~$2,499 | Qwen3.6 35B-A3B | 105 tok/s |
| Apple Mac Studio (M5 Max, 40-core GPU)Apple unified memory | 96 GBassumed | 614 GB/s | ~$3,099–$5,099 | Nemotron 3 Super 120B-A12B | 132 tok/s |
| Apple Mac Studio (M5 Ultra, 64-core GPU)Apple unified memory | 192 GBassumed | 1,200 GB/s | ~$5,499–$9,499 | GLM-5.3 Flash | 171 tok/s |
| Apple Mac Studio (M5 Ultra, 80-core GPU)Apple unified memory | 192 GBassumed | 1,200 GB/s | ~$6,799–$10,799 | GLM-5.3 Flash | ~200–357 tok/sestimate |
| Apple Mac mini (M5 Pro, 16-core GPU)Apple unified memory | 48 GBassumed | 307 GB/s | ~$1,699–$2,699 | Qwen3 Coder Next | 80 tok/s |
| Apple Mac mini (M5 Pro, 20-core GPU)Apple unified memory | 48 GBassumed | 307 GB/s | ~$1,899–$2,899 | Qwen3 Coder Next | 80 tok/s |
| Apple MacBook Pro 14-inch (M5 Max, 32-core GPU)Apple unified memory | 24 GBassumed | 460 GB/s | ~$4,099 | Qwen3.6 35B-A3B | 105 tok/s |
| Apple MacBook Pro 14-inch (M5 Max, 40-core GPU)Apple unified memory | 96 GBassumed | 614 GB/s | ~$4,699–$6,699 | Nemotron 3 Super 120B-A12B | 132 tok/s |
| Apple MacBook Pro 16-inch (M5 Max, 32-core GPU)Apple unified memory | 24 GBassumed | 460 GB/s | ~$4,399 | Qwen3.6 35B-A3B | 105 tok/s |
| Apple MacBook Pro 16-inch (M5 Max, 40-core GPU)Apple unified memory | 96 GBassumed | 614 GB/s | ~$4,999–$6,999 | Nemotron 3 Super 120B-A12B | 132 tok/s |