measuredlocally.comMachine Updated 20 October 2026
Apple Mac mini M4
Apple's Mac mini with the M4 chip and 16GB of memory shared between the processor and the graphics.
Measured on mine measured by me
| Set-up | Model | Writing code | Writing prose | Reading (prefill) | Tested |
|---|---|---|---|---|---|
| Mac mini M4llama.cpp on Metal | Gemma 4 12BGoogle's own 4-bit QAT file (Q4_0), GGUF | 12.9 tok/s | 13.3 tok/s | 115 tok/s | 20 Oct 2026 |
| Mac mini M4MLX (mlx-lm) on Metal | Qwen 3.5 9BMLX 4-bit | 20.5 tok/s | 20.6 tok/s | 207 tok/s | 20 Oct 2026 |
| Mac mini M4llama.cpp on Metal | Qwen 3.5 9BQ4_K_M, GGUF | 17.7 tok/s | 18.0 tok/s | 185 tok/s | 20 Oct 2026 |
| Mac mini M4Prism ML's llama.cpp on Metal | Ternary Bonsai 2 27BPTQ1_0, about 1.75 bits a weight, GGUF | 9.4 tok/s | 9.6 tok/s | 47 tok/s | 20 Oct 2026 |
One user, the model's thinking switched off, the same prompts and the same test each time. All of my results, and how I test, are on the hardware leaderboard.
The hardware, exactly
- Machine: Mac mini (2024).
- Chip: Apple M4: 10 processor cores (4 performance, 6 efficiency) and 10 graphics cores.
- Memory: 16GB, shared between the processor and the graphics, with a bandwidth of 120GB a second (Apple's figure).
- System: macOS 26.5.2.