How fast is Mistral Small 3 24B on an Mac?

Around 8 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 150 GB/s.

How much memory does Mistral Small 3 24B need?

13.5 GiB for the weights at Q4_K_M, plus 1.3 GiB for an 8K-token KV cache and about 1.5 GiB of runtime overhead — 16.2 GiB in total.

Home › Mistral Small 3 24B › M3 Pro 18GB

Can an M3 Pro 18GB Mac run Mistral Small 3 24B?

M3 Pro · 18 GB unified memory · 150 GB/s · MacBook Pro 14" (M3 Pro), MacBook Pro 16" (M3 Pro)

No — not at Q4_K_M.

Mistral Small 3 24B needs about 16.2 GiB but an M3 Pro 18GB Mac has only ~15.0 GiB available for a model — you are 1.2 GiB short. macOS will swap to SSD and generation will crawl. There is a smaller quantisation that fits: see below.

Weights

13.5 GiB

23.6B params @ Q4_K_M

KV cache

1.3 GiB

8K context, FP16

Overhead

1.5 GiB

runtime + activations

Total needed

16.2 GiB

of ~15.0 GiB usable

Est. speed

8 tok/s

150 GB/s bandwidth

Max context

1K 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	44.0 GiB	48.2 GiB	3	No
Q8_0	23.4 GiB	26.6 GiB	5	No
Q6_K	18.1 GiB	21.1 GiB	6	No
Q5_K_M	15.7 GiB	18.5 GiB	7	No
Q4_K_M	13.5 GiB	16.2 GiB	8	No
Q3_K_M	10.7 GiB	13.3 GiB	10	Yes, but tight
Q2_K	8.2 GiB	10.7 GiB	14	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	15.0 GiB	No	Yes, but tight
4K	0.6 GiB	15.4 GiB	No	No
8K	1.3 GiB	16.2 GiB	No	No
16K	2.5 GiB	17.7 GiB	No	No
32K	5.0 GiB	20.8 GiB	No	No

What to run instead on an M3 Pro 18GB Mac

These are the largest models that fit comfortably on this machine at Q4_K_M with an 8K context.

Model	Size	Total	Tok/s
DeepSeek-R1-Distill-Qwen-14B	14.8B	11.2 GiB	13	Yes, comfortably
Qwen2.5 14B	14.8B	11.2 GiB	13	Yes, comfortably
Qwen2.5-Coder 14B	14.8B	11.2 GiB	13	Yes, comfortably
Qwen3 14B	14.8B	10.9 GiB	13	Yes, comfortably
Phi-4 14B	14.7B	11.2 GiB	13	Yes, comfortably
Gemma 3 12B	12.2B	11.1 GiB	16	Yes, comfortably

Macs that run Mistral Small 3 24B comfortably

The smallest configuration that handles it is the M2 24GB (MacBook Air 15" (M2)).

Mac	Memory	Bandwidth	Tok/s
M2 24GB	24 GB	100 GB/s	6
M4 24GB	24 GB	120 GB/s	7
M4 Pro 24GB	24 GB	273 GB/s	15
M1 Max 32GB	32 GB	400 GB/s	22
M2 Max 32GB	32 GB	400 GB/s	22
M4 32GB	32 GB	120 GB/s	7
M3 Max 36GB	36 GB	300 GB/s	17
M4 Max 36GB	36 GB	410 GB/s	23
M4 Pro 48GB	48 GB	273 GB/s	15
M4 Max 48GB	48 GB	546 GB/s	30
M1 Max 64GB	64 GB	400 GB/s	22
M2 Ultra 64GB	64 GB	800 GB/s	44

Common questions

Can an M3 Pro 18GB Mac run Mistral Small 3 24B?: No. At Q4_K_M it needs about 16.2 GiB, and an M3 Pro 18GB Mac has roughly 15.0 GiB available for a model after macOS takes its share.
How fast is Mistral Small 3 24B on an Mac?: Around 8 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 150 GB/s.
How much memory does Mistral Small 3 24B need?: 13.5 GiB for the weights at Q4_K_M, plus 1.3 GiB for an 8K-token KV cache and about 1.5 GiB of runtime overhead — 16.2 GiB in total.