AppleMac / unified memoryCurrent

Apple M4 Pro for local LLMs

Name: LLM Configurator — GPU VRAM Checker
Author: LLM Configurator

Written by Jakub Rusinowski · Last updated July 12, 2026

With 24 GB at 273 GB/s, the M4 Pro runs 93 catalogued models at Q4_K_M with 8K context. The largest that fits is Qwen 3.6 27B (~17.6 GB), and the top pick is LFM2.5-24B-A2B at 46–96 tok/s.

24–48 GB unified memory covers most everyday AI tasks. 30W total system power. Ideal for laptop users who want serious AI without a desktop GPU.

Check my hardware →Full benchmark table

Models that run on the M4 Pro

Q4_K_M, 8K context, 18 GB usable. Ranked by quality and speed.

Model	VRAM	Speed	Fit
LFM2.5-24B-A2B LFM2.5	15.5 GB	46–96 tok/s	Fits
GPT-OSS 20B GPT-OSS	13.8 GB	32–67 tok/s	Fits
Cosmos 3 Nano Cosmos 3	10.5 GB	14–30 tok/s	Fits
Mellum2.1 12B Thinking Mellum2.1	9 GB	47–97 tok/s	Fits
Bielik PL 11B v3.0 Instruct Bielik	7.4 GB	12–24 tok/s	Fits
Llama 3.2 Vision 11B Llama 3.2 Vision	8.5 GB	12–25 tok/s	Fits
Falcon 3 10B Instruct Falcon 3	8.4 GB	12–25 tok/s	Fits
Qwen 3.5 9B Qwen 3.5	6.5 GB	16–33 tok/s	Fits
Gemma 4 12B (Unified) Gemma 4	8 GB	11–22 tok/s	Fits
Gemma 4 E2B Gemma 4	3.9 GB	21–45 tok/s	Fits
Gemma 4 E4B Gemma 4	5.6 GB	15–31 tok/s	Fits
Qwen 3 8B Qwen 3	7 GB	15–31 tok/s	Fits

Showing 12 of 93

Buy it or rent the same memory

Buy the card, or rent a GPU with the same memory by the hour to try models first.

Buy this hardware Apple Mac Mini M4 Pro — 64 GB unified memory · 273 GB/s Deploy in the cloud now RTX 4090 on RunPod — from $0.34/hr · rate checked 2026-07

or compare on Vast.ai from $0.35/hr (typical low · varies)

As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.

Affiliate disclosure: Some links on this page are affiliate links — if you buy through them, LLM Configurator may earn a commission at no extra cost to you. As an Amazon Associate, LLM Configurator earns from qualifying purchases.

Apple Mac Mini M4 Pro

64 GB unified memory · 273 GB/s

2026 prices are volatile — check the current listing.

Check price on Amazon

Speed vs other GPUs

Llama 3.1 8B, Q4_K_M. Estimated ranges. How this is calculated

RTX 5090113–234 tok/s RTX 409086–165 tok/s RTX 508074–153 tok/s RTX 5090 Laptop GPU70–145 tok/s RTX 5070 Ti70–145 tok/s RTX PRO 4500 Blackwell (Workstation Edition)70–145 tok/s RTX 5080 Laptop GPU70–145 tok/s RTX PRO 5000 Blackwell Laptop GPU70–145 tok/s M4 Pro17–35 tok/s

Specifications

Specs last updated 2026-07-12.

Memory: 24 GB
Memory bandwidth: 273 GB/s
Architecture: ARM, 3nm TSMC
Series: Apple Silicon
Board power: 30 W
Release year: 2024
Usable for models: 18 GB (~75% of unified)

Best for8B–14B modelsMacBook Proenergy efficiency

Recommended first model (Ollama)

Frequently asked questions

Can the Apple M4 Pro run local LLMs?

Yes. With 24 GB (18 GB usable by a model) it runs 93 of the catalogued models at Q4_K_M with 8K context; the largest is Qwen 3.6 27B, needing about 17.6 GB.

How fast is the Apple M4 Pro for AI inference?

It is estimated to run Llama 3.1 8B at 17–35 tok/s at Q4_K_M. Llama 3.3 70B does not fit: it needs about 44 GB against 18 GB usable. These are modelled estimates from memory bandwidth, not measurements; the methodology page shows the formula.

What LLMs can I run on 24 GB?

Among the best that fit: LFM2.5-24B-A2B, GPT-OSS 20B, Cosmos 3 Nano, Mellum2.1 12B Thinking, Bielik PL 11B v3.0 Instruct. The quickest start is Ollama: ollama run qwen3:14b.