Can an Apple M4 16GB run Qwen3.5 27B?

Q: Can an Apple M4 16GB run Qwen3.5 27B?

Yes, with a setting. It needs 13.7 GB at Q3_K_M and macOS allows 10.72 GB by default. Raise the limit and it runs at about 6.4 tokens per second.

In practice

On paper the Apple M4 16GB has 16 GB, and Qwen3.5 27B at Q3_K_M needs 13.7 GB. macOS does not let the GPU use all of it: by default about 67% is available, which is 10.72 GB. So the model is just over the default limit and well inside the physical memory.

The limit is a setting, and one command raises it for the current session. It resets at reboot, and the memory it takes comes from the rest of the system, so keep other large programs closed while the model is loaded. Once it is raised the estimate is about 6.4 tokens per second.

The numbers

Qwen3.5 27B has 27.78 billion parameters across 64 layers. The Apple M4 16GB has 16 GB of unified memory at 120 GB/s, of which about 10.72 GB is left for a model once the operating system has taken its share.

Quantization	Total needed	Fits	Speed
Q8_0	28.54 GB	no	n/a
Q6_K	22.27 GB	no	n/a
Q5_K_M	19.46 GB	no	n/a
Q4_K_M	16.67 GB	no	n/a
IQ4_XS	14.8 GB	no	n/a
Q3_K_M	13.7 GB	with raised limit	6.4 t/s
IQ3_XXS degraded	10.95 GB	with raised limit	8.1 t/s
Q2_K degraded	11.89 GB	with raised limit	7.4 t/s
IQ2_XXS degraded	7.72 GB	yes	11.7 t/s

1 of the 9 levels fit entirely. Levels marked degraded are listed for completeness, not as advice.

The settings we would use

sudo sysctl iogpu.wired_limit_mb=15360

macOS lets the GPU wire down about 67% of the memory by default, which is less than this model needs. The command raises the limit for this session and resets at reboot.

Once the limit is raised, Ollama and LM Studio use the extra memory without further settings. Run ollama ps and check that the processor column reads 100% GPU, and see the Mac and PC comparison for what unified memory does well and where it is slow.

If you would rather not change the limit

A smaller model from the same family is the straightforward answer. These fit on an Apple M4 16GB at a quantization worth using:

Qwen3 14B, 14.77B at IQ4_XS, 8.97 GB, about 10.0 tokens/s
Qwen2.5 14B Instruct, 14.77B at Q3_K_M, 8.63 GB, about 10.4 tokens/s
Qwen2.5 Coder 14B Instruct, 14.77B at Q3_K_M, 8.63 GB, about 10.4 tokens/s

What else an Apple M4 16GB runs well

For comparison, these are the other popular models that fit this device at a sound quantization, best pairings first.

Model	Quantization	Memory	Speed
Qwen3.5 9B	Q6_K	8.04 GB	11.8 t/s
Gemma 4 12B Instruct	Q4_K_M	8.41 GB	10.2 t/s
Qwen3 8B	Q6_K	7.76 GB	11.8 t/s
Qwen3 14B	IQ4_XS	8.97 GB	10.0 t/s

The same model on nearby hardware

Memory decides whether Qwen3.5 27B loads and bandwidth decides how quickly it answers. Here is what a step down, a card with the same memory and a step up do with it.

Device	Memory	Bandwidth	Result
GeForce RTX 4070	12 GB	504 GB/s	does not fit
GeForce RTX 5080	16 GB	960 GB/s	IQ4_XS, 51.3 t/s
GeForce RTX 4090	24 GB	1008 GB/s	Q6_K, 35.5 t/s

See how fast it feels

Apple M4 16GB running Qwen3.5 27B at Q3_K_M

Wait for the first word20 s

Then writes at6.4 tok/s

Whole answer48 s

YouWhy does my model use more memory when the conversation gets longer?

Model

Simulated from our estimate at a 512-token question, not a recording. Assumes nothing else is competing for the GPU.

Questions people ask

How much VRAM does Qwen3.5 27B need on Apple M4 16GB?

At Q4_K_M with an 8k context window it needs 16.67 GB in total: 15.62 GB of weights, 0.64 GB of KV cache and 0.41 GB of runtime overhead. The Apple M4 16GB makes about 10.72 GB available once the operating system has taken its share.

Can I run Qwen3.5 27B on Apple M4 16GB anyway?

Yes, after raising the macOS GPU memory limit, which one sysctl command does. Without it the model needs 13.7 GB and the default limit allows 10.72 GB.

Keep going

Everything the Apple M4 16GB runs, the full breakdown for Qwen3.5 27B, how the numbers are made in the methodology, or a different pairing in the calculator.