How fast is Qwen2.5 72B on an Mac?

Around 15 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 819 GB/s.

How much memory does Qwen2.5 72B need?

41.5 GiB for the weights at Q4_K_M, plus 2.5 GiB for an 8K-token KV cache and about 2.9 GiB of runtime overhead — 46.8 GiB in total.

Home › Qwen2.5 72B › M3 Ultra 96GB

Can an M3 Ultra 96GB Mac run Qwen2.5 72B?

M3 Ultra · 96 GB unified memory · 819 GB/s · Mac Studio (M3 Ultra)

Yes — Qwen2.5 72B runs on an M3 Ultra 96GB Mac.

At Q4_K_M it needs about 46.8 GiB, leaving 34.8 GiB spare out of the ~81.6 GiB macOS will let you use. Expect roughly 15 tokens/sec — slow but workable for background tasks.

Weights

41.5 GiB

72.7B params @ Q4_K_M

KV cache

2.5 GiB

8K context, FP16

Overhead

2.9 GiB

runtime + activations

Total needed

46.8 GiB

of ~81.6 GiB usable

Est. speed

15 tok/s

819 GB/s bandwidth

Max context

64K 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	135.4 GiB	145.5 GiB	5	No
Q8_0	71.9 GiB	78.8 GiB	8	Yes, but tight
Q6_K	55.9 GiB	62.0 GiB	11	Yes, comfortably
Q5_K_M	48.2 GiB	54.0 GiB	13	Yes, comfortably
Q4_K_M	41.5 GiB	46.8 GiB	15	Yes, comfortably
Q3_K_M	33.0 GiB	38.0 GiB	18	Yes, comfortably
Q2_K	25.4 GiB	30.0 GiB	24	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	0.6 GiB	44.7 GiB	Yes, comfortably	Yes, comfortably
4K	1.3 GiB	45.4 GiB	Yes, comfortably	Yes, comfortably
8K	2.5 GiB	46.8 GiB	Yes, comfortably	Yes, comfortably
16K	5.0 GiB	49.6 GiB	Yes, comfortably	Yes, comfortably
32K	10.0 GiB	55.2 GiB	Yes, comfortably	Yes, comfortably
128K	40.0 GiB	88.8 GiB	No	Yes, but tight

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 qwen2.5:72b

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/Qwen2-5-72B-4bit --prompt "Hello"

Macs that run Qwen2.5 72B comfortably

The smallest configuration that handles it is the M2 Max 96GB (MacBook Pro 16" (M2 Max)).

Mac	Memory	Bandwidth	Tok/s
M2 Max 96GB	96 GB	400 GB/s	7
M3 Ultra 96GB	96 GB	819 GB/s	15
M3 Max 128GB	128 GB	400 GB/s	7
M4 Max 128GB	128 GB	546 GB/s	10
M2 Ultra 192GB	192 GB	800 GB/s	14
M3 Ultra 256GB	256 GB	819 GB/s	15
M3 Ultra 512GB	512 GB	819 GB/s	15

Common questions

Can an M3 Ultra 96GB Mac run Qwen2.5 72B?: Yes. At Q4_K_M it needs about 46.8 GiB of the roughly 81.6 GiB available, generating around 15 tokens per second.
How fast is Qwen2.5 72B on an Mac?: Around 15 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 819 GB/s.
How much memory does Qwen2.5 72B need?: 41.5 GiB for the weights at Q4_K_M, plus 2.5 GiB for an 8K-token KV cache and about 2.9 GiB of runtime overhead — 46.8 GiB in total.