Skip to content

Qwen3.6 27B on Mac mini M4 32 GB: VRAM, speed and fit

Qwen3.6 27B at 4-bit runs on the M4 32 GB with 56K of context at about 5.9 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on M4 32 GB · 4-bit (Q4_K_M) · Ollama / LM Studio · Chat

GPU memory
30.9 GB
of 32.0 GB
Context
56K
auto
Speed
5.9 tok/s
4.8 when full
First reply
35 s
2K prompt

Where the memory goes

Model
15.7 GB
Context
3.6 GB
Overhead
901 MB
OS
10.7 GB
Free
1.1 GB
Memory
30.9 GB of 32.0 GB
Context
56K tokens · auto
Speed
5.9 tok/s · 4.8 when full
First reply
35 s for a 2K prompt
Running cost
$0.85/mo at home · $33/mo on a cloud RTX A5000

Validation: Not validated

What Qwen3.6 27B on Mac mini M4 32 GB was checked against
CheckTargetResultStatus
Fits in GPU memory21.3 GB usable20.2 GB neededPass
Headroom5% free1.1 GB free (5%)Pass
Context for chat16K or more56KPass
Writing speed15 tok/s or more5.9 tok/sFails
First replyunder 3 s, 2K prompt35 sFails

Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.

The setup

Model
Qwen3.6 27B
Parameters
27.8B
Precision
4-bit (Q4_K_M)
Context
56K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
M4 32 GB
GPU can use
21.3 GB of 32.0 GB

Worth trying

  • 8-bit cache: Room for 104K context

Running cost

At home
$0.85/mo
39 W at the wall
Renting instead
$33/mo
RTX A5000 at $0.27/h

4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 22, 2026.

How to run it

llama.cpp
llama-server -hf prism-ml/Ternary-Bonsai-27B-gguf
vLLM
vllm serve Qwen/Qwen3.6-27B
MLX (Mac)
mlx_lm.chat --model mlx-community/Qwen3.6-27B-4bit

Qwen3.6 27B on similar machines

All machines: Qwen3.6 27B VRAM requirements. Everything that runs here: what can the Mac mini M4 32 GB run?

Other models that run on Mac mini M4 32 GB

Questions

Can Qwen3.6 27B run on Mac mini M4 32 GB?

Qwen3.6 27B at 4-bit runs on the M4 32 GB with 56K of context at about 5.9 tokens a second.

Report VR-1M0WBYA, calculated Nov 22, 2026 from published specs, not benchmarked.