Apple · M4 Pro · Mini PC
Best AI Models for Mac Mini M4 Pro (64 GB)
Runs 14B and 35B MoE models well and fits 120B MoE; 27–31B and 70B fit but are slow (11 and 4 tok/s).
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB | 43.2 tok/s | |
| 2 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB | 49.1 tok/s | |
| 3 | Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB | 49.1 tok/s | |
| 4 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 54.5 tok/s | |
| 5 | Q4_K_M · 19.3 GB | 55.7 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB | 43.2 tok/s | |
| 2 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB | 49.1 tok/s | |
| 3 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 54.5 tok/s | |
| 4 | Q4_K_M · 7.9 GB | 36.1 tok/s | |
| 5 | Q4_K_M · 19.3 GB | 55.7 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB | 43.2 tok/s | |
| 2 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB | 49.1 tok/s | |
| 3 | Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB | 30.0 tok/s | |
| 4 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB | 54.5 tok/s | |
| 5 | Q4_K_M · 19.3 GB | 55.7 tok/s |
| # | Model | Speed | Score |
|---|---|---|---|
| 1 | Qwen3.6 35B A3BAlibabaHow to run → Q4_K_M · 22.0 GB | 43.2 tok/s | |
| 2 | Qwen3.8 27BAlibaba Q4_K_M · 17.9 GB | 11.0 tok/s | |
| 3 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB | 11.0 tok/s | |
| 4 | Q4_K_M · 19.3 GB | 55.7 tok/s | |
| 5 | GLM 4.7 FlashZ.ai Q4_K_M · 19.9 GB | 40.5 tok/s |
| # | Model | Speed ↓ | Score |
|---|---|---|---|
| 1 | GPT OSS 20BOpenAIHow to run → Q4_K_M · 13.1 GB | 54.5 tok/s | |
| 2 | Q4_K_M · 19.3 GB | 55.7 tok/s | |
| 3 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB | 49.1 tok/s | |
| 4 | Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB | 49.1 tok/s | |
| 5 | Qwen3 30B A3B Instruct 2507Alibaba Q4_K_M · 19.4 GB | 49.1 tok/s |
The Mac Mini M4 Pro is a 64 GB mini PC with more memory than speed: it runs 14B and 35B MoE models well and fits 120B MoE. macOS keeps about 3.5 GB, leaving 60.5 GB for models. Prompt reading is its weak spot: a 4K-token prompt waits 11 s before the first word on Qwen3.6 35B A3B. M4 Pro generation, released 2024; the newer M5 Pro writes 13% faster with 1.9× the compute.
Mac Mini M4 Pro strengths and weaknesses
Strengths
- Runs 120B MoE models: GPT OSS 120B at 44.7 tok/s in 59.1 GB
Weaknesses
- Slow on large dense models: Llama 3.3 70B Instruct at 4.1 tok/s, Qwen3.8 27B at 11.0 tok/s
- Slow prompt reading: A 32K-token prompt on Qwen3.6 35B A3B waits 1.5 min for the first token
- No CUDA: vLLM, AWQ, GPTQ and FP8 builds don't run; use GGUF or MLX
Memory for models
60.5 GB of 64 GBEverything on the size ladder fits. Qwen3.6 35B A3B and Gemma 4 26B A4B IT together use 38.5 GB and leave 22 GB free.
Other memory sizes of the same machine first, then machines and GPUs people weigh against it, each with the one reason it is worth a look. Speeds on Qwen3.5 9B.
Output speed depends on memory bandwidth (273 GB/s). Grey rows need more than the 60.5 GB available.
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB30.0tok/s
- 14BQwen3 14B · Q4_K_M · 11.9 GB20.1tok/s
- 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB11.0tok/s
- 35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB43.2tok/s
- 4.1tok/s
- 44.7tok/s
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
Pro generations: writing vs compute
M1 Pro → M5 ProWriting speed follows memory bandwidth and barely moved: 22.0 to 33.8 tok/s since the M1 Pro. Compute, which sets how fast prompts are read, grew 3.1×. Indexed to the M1 Pro; speeds on Qwen3.5 9B, same 32 GB.
All 1168 models on the Mac Mini M4 Pro
Featured onlyBest quant that fits in 60.5 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
Nemotron H 8B Reasoning 128KChatReasoning | BF16 | 20.4 GB | 10.7 t/s | — | B49, Decent | |
| Q4_K_M | 6.2 GB | 35.5 t/s | 8K | A73, Runs well | ||
| Q4_K_M | 51.3 GB | 4.1 t/s | — | C34, Usable | ||
| BF16 | 17.9 GB | 10.9 t/s | 4K | B49, Decent | ||
Hermes 2 Theta Llama 3 70BChatFunctions | Q4_K_M | 45.3 GB | 4.4 t/s | 8K | C41, Usable | |
Qwen 72BChat | Q4_K_M | 52.5 GB | 4.0 t/s | 33K | C31, Usable | |
Meta Llama 3 70BChat | Q4_K_M | 51.3 GB | 4.1 t/s | — | C34, Usable | |
Llama 2 7B HFChat | Q4_K_M | 6.5 GB | 35.3 t/s | 4K | A73, Runs well | |
| BF16 | 18.5 GB | 11.9 t/s | — | B51, Decent | ||
DeepSeek Coder v2 Lite BaseChatCode | Q4_K_M | 10.2 GB | 69.3 t/s | 164K | S86, Runs great | |
Qwen1.5 32B ChatChat | Q4_K_M | 24.1 GB | 9.4 t/s | 33K | B47, Decent | |
Codegemma 7B ITChatCode | Q4_K_M | 8.2 GB | 33.9 t/s | — | A73, Runs well | |
QwQ 32B PreviewChatReasoning | Q4_K_M | 24.3 GB | 9.3 t/s | 33K | B47, Decent | |
| Q4_K_M | 6.5 GB | 34.7 t/s | 4K | A73, Runs well | ||
| Q3_K_M | 59.9 GB | 3.5 t/s | 33K | F10, Won't fit |
Specs
- Chip
- M4 Pro
- CPU cores
- 14
- GPU cores
- 20
- Unified memory
- 64 GB
- For models
- 60.5 GB (macOS keeps 3.5 GB)
- Memory bandwidth
- 273 GB/s
- FP16 (dense)
- 17.2 TFLOPS
- Neural Engine
- 38 TOPS
- Released
- 2024-11-08
Mac Mini M4 Pro FAQ
How much memory is available for AI on the Mac Mini M4 Pro (64 GB)?
About 60.5 GB of the 64 GB. macOS keeps roughly 3.5 GB for itself and other apps.
What is the largest model the Mac Mini M4 Pro (64 GB) can run?
GPT OSS 120B (116.8B) at Q4_0 uses 59.1 GB, 98% of the available memory, and runs at about 44.7 tok/s.
Can the Mac Mini M4 Pro (64 GB) run Qwen3.6 35B A3B?
Yes. Qwen3.6 35B A3B fits at Q4_K_M, using 22.0 GB of the 60.5 GB available, and it is the best model that runs well on this machine.
Can the Mac Mini M4 Pro (64 GB) run 70B models?
Yes, but slowly: Llama 3.3 70B Instruct at Q4_K_M uses 51.3 GB and writes about 4.1 tok/s.
How long does the Mac Mini M4 Pro (64 GB) take to read a long prompt?
On Qwen3.6 35B A3B, about 11 s for a 4K-token prompt and 1.5 min for a 32K-token document before the first word appears.
Is renting a GPU cheaper than buying the Mac Mini M4 Pro?
An A100 (80 GB) rents from $1.07/hr.