Mini
Apple Mac mini (M4, 16GB)
macOS
M4 with 16GB unified memory (512GB SSD baseline).
Memory
16 GB
GPUs
1×
RAM
16 GB
Models
24
Type
Mini
Inference Memory
16 GB
Accelerator
16 GB
System RAM
16 GB
OS
macOS
What it can run
24 models| Arcee Trinity Mini 26B | Q3_K_M | 8 tok/s | Not viable |
| Arcee Trinity Nano 6B | Q8_0 | 26 tok/s | Good |
| DeepSeek V3 | Q2_K | – | Not viable |
| Gemma 3 27B | Q4_K_M | – | Not viable |
| Gemma 3 4B | Q5_K_M | 19 tok/s | Good |
| Gemma 4 26B-A4B | Q3_K_M | 8 tok/s | Not viable |
| Gemma 4 31B | Q3_K_M | 1 tok/s | Not viable |
| Gemma 4 E2B | Q8_0 | 18 tok/s | Acceptable |
| Gemma 4 E4B | Q8_0 | 11 tok/s | Marginal |
| GigaChat Lightning 10B | Q8_0 | 44 tok/s | Acceptable |
| Llama 3.1 8B Instruct | Q8_0 | 16 tok/s | Good |
| Llama 3.2 11B Vision | Q8_0 | 14 tok/s | Good |
| Llama 3.2 1B Instruct | Q8_0 | 45 tok/s | Excellent |
| Llama 3.2 3B Instruct | Q8_0 | 30 tok/s | Excellent |
| NVIDIA Nemotron-3-super-120B-A12B | Q2_K | – | Not viable |
| Phi-4 Mini | Q8_0 | 28 tok/s | Good |
| Qwen 2.5 Coder 32B Instruct | Q4_K_M | – | Not viable |
| Qwen 2.5 Coder 7B Instruct | Q5_K_M | 18 tok/s | Good |
| Qwen3.5-122B-A10B | Q3_K_M | – | Not viable |
| Qwen3.5-27B | Q3_K_M | 20 tok/s | Acceptable |
| Qwen3.5-397B (MoE) | Q2_K | – | Not viable |
| Qwen3.6-27B | Q3_K_M | 20 tok/s | Acceptable |
| Stable Diffusion 3.5 Large | FP16 | – | Acceptable |
| Whisper Large V3 Turbo | FP16 | – | Good |
Showing 24 of 24 entries
Best Fit
Who this machine makes sense for
This machine is a buy-it-ready path for users who want predictable local AI performance without building from parts. 16 GB gives it enough headroom to matter for real model selection, not just toy workloads.
Before You Buy
What to verify first
The main check before buying is upgrade path clarity: confirm memory ceiling, storage expandability, and whether the accelerator path still matches the models you expect to run a year from now.