How fast is Qwen3 14B on an Mac?

Around 18 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 200 GB/s.

How much memory does Qwen3 14B need?

8.4 GiB for the weights at Q4_K_M, plus 1.3 GiB for an 8K-token KV cache and about 1.2 GiB of runtime overhead — 10.9 GiB in total.

Home › Qwen3 14B › M1 Pro 16GB

Can an M1 Pro 16GB Mac run Qwen3 14B?

M1 Pro · 16 GB unified memory · 200 GB/s · MacBook Pro 14" (M1 Pro), MacBook Pro 16" (M1 Pro)

Yes — Qwen3 14B runs on an M1 Pro 16GB Mac.

At Q4_K_M it needs about 10.9 GiB, leaving 2.1 GiB spare out of the ~13.0 GiB macOS will let you use. Expect roughly 18 tokens/sec — comfortable for interactive chat.

Weights

8.4 GiB

14.8B params @ Q4_K_M

KV cache

1.3 GiB

8K context, FP16

Overhead

1.2 GiB

runtime + activations

Total needed

10.9 GiB

of ~13.0 GiB usable

Est. speed

18 tok/s

200 GB/s bandwidth

Max context

16K tokens

at Q4_K_M

Where the memory goes

Three things occupy memory when a model is loaded: the weights themselves, the KV cache that holds the conversation, and the runtime's own working buffers. Only the first is fixed — the KV cache grows with every token in your context.

Quantisation	Weights	Total	Tok/s	Fits?
FP16	27.6 GiB	31.0 GiB	5	No
Q8_0	14.6 GiB	17.4 GiB	10	No
Q6_K	11.4 GiB	14.0 GiB	13	No
Q5_K_M	9.8 GiB	12.4 GiB	15	Yes, but tight
Q4_K_M	8.4 GiB	10.9 GiB	18	Yes, comfortably
Q3_K_M	6.7 GiB	9.1 GiB	22	Yes, comfortably
Q2_K	5.2 GiB	7.5 GiB	29	Yes, comfortably

The default. Best quality-per-gigabyte for most people.

How long a context fits

The KV cache is often what breaks a setup that looked fine at load time. Quantising it to 8-bit roughly halves its footprint with little measurable quality cost.

Context	KV cache	Total	FP16 KV	Q8 KV
2K	0.3 GiB	9.8 GiB	Yes, comfortably	Yes, comfortably
4K	0.6 GiB	10.1 GiB	Yes, comfortably	Yes, comfortably
8K	1.3 GiB	10.9 GiB	Yes, comfortably	Yes, comfortably
16K	2.5 GiB	12.5 GiB	Yes, but tight	Yes, but tight
32K	5.0 GiB	15.6 GiB	No	No
128K	20.0 GiB	34.2 GiB	No	No

How to run it

Ollama is the shortest path. It picks a quantisation automatically — usually Q4_K_M, which is what the numbers above assume.

brew install ollama
ollama serve &
ollama run qwen3:14b

For MLX — Apple's own array framework, typically a little faster than llama.cpp on Apple Silicon:

pip install mlx-lm
mlx_lm.generate --model mlx-community/Qwen3-14B-4bit --prompt "Hello"

Macs that run Qwen3 14B comfortably

The smallest configuration that handles it is the M1 16GB (MacBook Air (M1)).

Mac	Memory	Bandwidth	Tok/s
M1 16GB	16 GB	68 GB/s	6
M1 Pro 16GB	16 GB	200 GB/s	18
M2 16GB	16 GB	100 GB/s	9
M4 16GB	16 GB	120 GB/s	11
M3 Pro 18GB	18 GB	150 GB/s	13
M2 24GB	24 GB	100 GB/s	9
M4 24GB	24 GB	120 GB/s	11
M4 Pro 24GB	24 GB	273 GB/s	24
M1 Max 32GB	32 GB	400 GB/s	35
M2 Max 32GB	32 GB	400 GB/s	35
M4 32GB	32 GB	120 GB/s	11
M3 Max 36GB	36 GB	300 GB/s	26

Common questions

Can an M1 Pro 16GB Mac run Qwen3 14B?: Yes. At Q4_K_M it needs about 10.9 GiB of the roughly 13.0 GiB available, generating around 18 tokens per second.
How fast is Qwen3 14B on an Mac?: Around 18 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 200 GB/s.
How much memory does Qwen3 14B need?: 8.4 GiB for the weights at Q4_K_M, plus 1.3 GiB for an 8K-token KV cache and about 1.2 GiB of runtime overhead — 10.9 GiB in total.