How fast is DeepSeek-R1-Distill-Llama-70B on an Mac?

Around 2 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 120 GB/s.

How much memory does DeepSeek-R1-Distill-Llama-70B need?

40.3 GiB for the weights at Q4_K_M, plus 2.5 GiB for an 8K-token KV cache and about 2.8 GiB of runtime overhead — 45.6 GiB in total.

Home › DeepSeek-R1-Distill-Llama-70B › M4 24GB

Can an M4 24GB Mac run DeepSeek-R1-Distill-Llama-70B?

M4 · 24 GB unified memory · 120 GB/s · MacBook Air 15" (M4), Mac mini (M4)

No — not at Q4_K_M.

DeepSeek-R1-Distill-Llama-70B needs about 45.6 GiB but an M4 24GB Mac has only ~20.4 GiB available for a model — you are 25.2 GiB short. macOS will swap to SSD and generation will crawl.

Weights

40.3 GiB

70.6B params @ Q4_K_M

KV cache

2.5 GiB

8K context, FP16

Overhead

2.8 GiB

runtime + activations

Total needed

45.6 GiB

of ~20.4 GiB usable

Est. speed

2 tok/s

120 GB/s bandwidth

Max context

n/a

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	131.5 GiB	141.4 GiB	1	No
Q8_0	69.9 GiB	76.7 GiB	1	No
Q6_K	54.2 GiB	60.3 GiB	2	No
Q5_K_M	46.8 GiB	52.5 GiB	2	No
Q4_K_M	40.3 GiB	45.6 GiB	2	No
Q3_K_M	32.1 GiB	37.0 GiB	3	No
Q2_K	24.7 GiB	29.2 GiB	4	No

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	43.5 GiB	No	No
4K	1.3 GiB	44.2 GiB	No	No
8K	2.5 GiB	45.6 GiB	No	No
16K	5.0 GiB	48.4 GiB	No	No
32K	10.0 GiB	54.0 GiB	No	No
128K	40.0 GiB	87.6 GiB	No	No

What to run instead on an M4 24GB 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
Mistral Small 3 24B	23.6B	16.2 GiB	7	Yes, comfortably
gpt-oss-20b	20.9B	13.7 GiB	27	Yes, comfortably
DeepSeek-R1-Distill-Qwen-14B	14.8B	11.2 GiB	11	Yes, comfortably
Qwen2.5 14B	14.8B	11.2 GiB	11	Yes, comfortably
Qwen2.5-Coder 14B	14.8B	11.2 GiB	11	Yes, comfortably
Qwen3 14B	14.8B	10.9 GiB	11	Yes, comfortably

Macs that run DeepSeek-R1-Distill-Llama-70B comfortably

The smallest configuration that handles it is the M1 Max 64GB (MacBook Pro 16" (M1 Max)).

Mac	Memory	Bandwidth	Tok/s
M1 Max 64GB	64 GB	400 GB/s	7
M2 Ultra 64GB	64 GB	800 GB/s	15
M4 Pro 64GB	64 GB	273 GB/s	5
M4 Max 64GB	64 GB	546 GB/s	10
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	15
M3 Ultra 256GB	256 GB	819 GB/s	15
M3 Ultra 512GB	512 GB	819 GB/s	15

Common questions

Can an M4 24GB Mac run DeepSeek-R1-Distill-Llama-70B?: No. At Q4_K_M it needs about 45.6 GiB, and an M4 24GB Mac has roughly 20.4 GiB available for a model after macOS takes its share.
How fast is DeepSeek-R1-Distill-Llama-70B on an Mac?: Around 2 tokens per second. Token generation on Apple Silicon is limited by memory bandwidth, and this machine has 120 GB/s.
How much memory does DeepSeek-R1-Distill-Llama-70B need?: 40.3 GiB for the weights at Q4_K_M, plus 2.5 GiB for an 8K-token KV cache and about 2.8 GiB of runtime overhead — 45.6 GiB in total.